Marcus L Endicott: Quora Answers



Topics: Apache Hadoop | APIs | Artificial General Intelligence | Artificial Intelligence | Artificial Intelligence Markup Language AIML | Artificial Intelligence Startups | Augmented Reality | Automatic Summarization | Bots | Chat Automation | Chatbots | Chatterbot | Classification machine learning | Cleverbot | Cloud Computing | Cognitive Science | Computational Dreaming | Computational Linguistics | Content Analysis | Conversational Agents Chatbots | Conversation Software | DataSift | Diffbot | Digital Smart Homes | Discourse Analysis | Discovery Engines | Domotics | Eastern Culture | Expert Systems | eXtensible Messaging and Presence Protocol XMPP | FeedBurner | Game based Learning | Google Reader | Home Automation | IBM Watson | ICQ | Imified | Instant Messaging IM | iPhone Applications | Machine Learning | Metaphors | Middleware | Mind Mapping | Natural Language Generation | Natural Language Processing | OPML | Paper li | Pattern Recognition | Personal Productivity Apps | Project Glass | Q A Websites | Question Answering System | Question Generation | Recommendation Engines | Robots | RSS | Semantic Web | Similar Companies to X | Siri software | Smart Home | Software as a Service SaaS | Speech Recognition | Speech to Text | Spell Checking | Stanford NLP | Startups in Poland | Suggestion Engines | Summarization | Summify | Text Analytics | Text Synthesis | Travel Startups and Companies | Turing Test | Twitter Bots | Twitter Retweets | uClassify | User Data | Virtual Assistants bots | Virtual Self | Wavii | Web Content Curation | Web Feeds | Web Scraping | Wolfram Alpha | Word Clouds | Yahoo! Pipes 

[124x Oct 2012]


Technology Trends: Who is creating AI task oriented online Agents for consumers?

See my recent Quora answer to: Which is the best siri like app for android?

>>There are basically two kinds of apps here: apps that will talk with you, and apps that will do things for you (so called "do-engines").  Maybe three kinds if you count apps that do both.  Some people are trying to refer to these personal assistant apps as "answer devices" (http://answerdevices.com/).<<

In terms of do-engines, in the beginning there were IRC bots, often erroneously referred to as "chatbots"; natural language commands do not a chatbot make.

Probably Google Now came first with the ambient push, then MindMeld came next with natural language anticipative contextualization, and the newcomer on this block may be Kimera Systems (@kimerasystems) Strong AI (artificial intelligence) by Mounir Shita [1].

[1] http://twitter.com/mendicot/status/258219828719005696
  
= = =

Conversational Agents (Chatbots): What are some good bots or chatbots online that reasonably approximate human conversation, or do the best job of doing so?

Of course, it depends on how you evaluate "good" or "human-like", and there is all kinds of debate around the various Turing Tests.  I can recommend that you begin with my Meta Guide webpages, "100 Best Chat Bot Videos" [1], "100 Best Chatbot Videos" [2], and "100 Best Chatbots Videos" [3].  (I can't help but feel that the number of independent videos made about an app reflects its relative popularity.)

[1] http://www.meta-guide.com/home/ai-engine/100-best-chat-bot-videos

[2] http://www.meta-guide.com/home/ai-engine/100-best-chatbot-videos

[3] http://www.meta-guide.com/home/ai-engine/100-best-chatbots-videos

See also:
- Which is the best siri like app for android?
  
= = =

How would you build a robot to answer questions on Quora?

See my recent Quora answers to:

- What are some creative ideas for Quora's (future) API?
- What are the key challenges in designing/developing an effective question answering system?
- Can computers make questions?
- Would you pay—real money, not credits—for a good answer?
- Are there any "mechanical turk" mobile apps for earning money on smartphones?
- Could game-based welfare be an answer to machine induced high unemployment?
  
= = =

Is there a way to automatically email each new entry of a RSS feed to your email address?

Lots, from my Meta Guide webpage, "Best Web 2.0 Tools" [1] :

http://www.blogtrottr.com
http://www.feedmailer.net
http://feedmailpro.com
http://www.feedmyinbox.com
http://www.quickthreads.com
http://www.rssforward.com

[1] http://www.meta-guide.com/home/about/best-web-2-0-tools
  
= = =

Machine Learning : What would be a good way to organize the field of machine learning into sub-fields, for a new research candidate to gain a good overview of the domain. Basically what are the different streams of research active currently?

I suspect that the Wikipedia "Category:Machine_learning" [1] is as good a breakdown as any; so, because I'm interested in all applications of machine learning to dialog systems, I applied that list of sub-topics to a consistent search of Google Scholar [2].  What I can see clearly from that exercise is which sub-topics are not covered at all (unlinked) as well as the exact number of hits for each sub-topic, which gives me a good indication of the relative popularity of the various machine learning sub-topics as applied to dialog systems.

[1] http://en.wikipedia.org/wiki/Category:Machine_learning

[2] http://www.meta-guide.com/home/bibliography/google-scholar/machine-learning-dialog-systems
  
= = =

Quora: What are some good ideas for features Quora can add for making the spending of Quora credits more fun?

It would be nice if Quora credits could be donated to support some charity of one's choice, at least contributing towards a solution for someone [1].

[1] Would you pay—real money, not credits—for a good answer?
  
= = =

Quora API: What are some creative ideas for Quora's (future) API?

Think App.net ... and make it very friendly for Quora Bots.

See also:
- What are the key challenges in designing/developing an effective question answering system?
- Can computers make questions?
- Would you pay—real money, not credits—for a good answer?
- Are there any "mechanical turk" mobile apps for earning money on smartphones?
- Could game-based welfare be an answer to machine induced high unemployment?
  
= = =

Smart Home: What are some of the leading companies developing smart home technologies?

Microsoft HomeOS [1] should be mentioned here.  There is a "List of home automation software" [2] maintained on Wikipedia.  I can also refer you to my own Meta Guide webpages, "Dialog Systems in Smart Home Automation & Domotics (Domestic Robotics)" [3] as well as "Best Internet of Things Resources" [4].

[1] http://en.wikipedia.org/wiki/HomeOS

[1] http://en.wikipedia.org/wiki/List_of_home_automation_software

[3] http://www.meta-guide.com/home/ai-engine/home-automation

[4] http://www.meta-guide.com/home/about/best-internet-of-things-resources
  
= = =

What are the most notable examples of expert systems that were used in space?

http://www.newscientist.com/article/dn7584

> Space station gets HAL-like computer

http://ti.arc.nasa.gov/tech/cas/user-centered-technologies/clarissa/

>>
Clarissa is a fully voice-operated procedure browser, enabling astronauts to be more efficient with their hands and eyes and to give full attention to the task while they navigate through the procedure using spoken commands. The software was installed on the ISS in January 2005, and was first used by Expedition 11 Science Officer and Flight Engineer John Phillips on June 27, 2005. To the best of our knowledge, this is the first ever use of a spoken dialogue system in space. During the test, Phillips completed the interactive Clarissa training procedure, which exercises all the main system functionality; this procedure contains 50 steps, and took 25 minutes to complete. Speech recognition and dialogue management functioned well. A report summarizing the results is available here.
<<
  
= = =

What features of Google Reader are most useful?

I love that you can get Google Alerts feeds in Google Reader.
I love that you can export and import OPML.
I love that you can get output feeds via public folders.

I hate that there are no filters.

My ideal solution would be to merge FeedBurner into Google Reader, and then mashup Yahoo! Pipes inside Google Reader.

- Why is there no plug and play database as a service for web feeds?

The other thing is that there is no plug and play DBaaS expressly for web feeds; I would love to be able to accumulate custom DBs from feed items, and then be able to easily query them via API with feed output.
  
= = =

Apple Inc.: Which are the sources of siri?

According to ProgrammableWeb, Apple Siri uses more than 35 APIs [1].  For instance, Siri accounts for about a quarter of the queries fielded by Wolfram|Alpha [2].  Yelp is another of the APIs Siri uses [3].

[1] http://blog.programmableweb.com/2011/10/04/are-there-open-apis-behind-apples-new-voice-commands/

[2] http://www.nytimes.com/2012/02/07/technology/wolfram-a-search-engine-finds-answers-within-itself.html

[3] http://officialblog.yelp.com/2012/03/your-iphone-personal-assistant-now-with-more-yelp.html

See also:
- Which topics on Quora are related to the technology behind Apple Siri?
- What software stack does Siri use?
  
= = =

Android Applications: Is there an app similar to the Siri app for the iPhone 4?

It depends on what you mean by "similar"; but, there are quite a few, not to mention Cydia jailbreak options.  Just to name some for iOS, there are Evi [1], Speaktoit [2], and Voice Actions [3].

[1] http://itunes.apple.com/us/app/evi/id463296609

[2] http://itunes.apple.com/us/app/speaktoit-assistant/id491854246

[3] http://itunes.apple.com/us/app/voice-actions/id422037126

If you mean Android apps similar to Siri, then there are even more options available.  I suggest checking out my Meta Guide webpages, "100 Best Android Assistant Videos" [4] and "Best Chatbot Apps" [5].

[4] http://www.meta-guide.com/home/about/best-of-the-best-videos/100-best-android-assistant-videos

[5] http://www.meta-guide.com/home/ai-engine/best-chatbot-apps

= = =

Comedy: What's the most hilarious thing Siri has said?

I've put together a webpage listing "100 Best Siri Joke Videos" [1].  BTW, Sam Joseph (@tansaku) is organizing a "Funniest Computer Ever" [2] competition.  You can read more about "Computational Humor" [3] on Wikipedia.

[1] http://www.meta-guide.com/home/about/best-of-the-best-videos/100-best-siri-joke-videos

[2] http://funniestcomputer.neurogrid.com/

[3] http://en.wikipedia.org/wiki/Computational_humor

= = =

Android (OS): Which is the best siri like app for android?

There are basically two kinds of apps here: apps that will talk with you, and apps that will do things for you (so called "do-engines").  Maybe three kinds if you count apps that do both.  Some people are trying to refer to these personal assistant apps as "answer devices" (http://answerdevices.com/).  I can't help but feel that the number of independent videos made about an app reflects its relative popularity; so, I've now got Meta Guide webpages for all of the personal assistant apps mentioned in this thread, and then some....

100 Best Google Now Videos
http://www.meta-guide.com/home/about/best-of-the-best-videos/100-best-google-now-videos

100 Best Iris App Videos
http://www.meta-guide.com/home/ai-engine/best-chatbot-apps/100-best-iris-app-videos

Best EVA Assistant Videos
http://www.meta-guide.com/home/about/best-of-the-best-videos/best-eva-assistant-videos

Best Magnifis Robin Videos
http://www.meta-guide.com/home/about/best-of-the-best-videos/best-magnifis-robin-videos

Best Maluuba App Videos
http://www.meta-guide.com/home/about/best-of-the-best-videos/best-maluuba-app-videos

Best Skyvi App Videos
http://www.meta-guide.com/home/ai-engine/best-chatbot-apps/best-skyvi-app-videos

Best Speaktoit App Videos
http://www.meta-guide.com/home/ai-engine/best-chatbot-apps/best-speaktoit-app-videos

Best Vlingo App Videos
http://www.meta-guide.com/home/ai-engine/best-chatbot-apps/best-vlingo-app-videos
= = =

User Behavior on Quora: What are some different philosophies with respect to how much time to spend on Quora's home page, as opposed to browsing, searching, and using other techniques to find content on the site?

IMHO, Quora search does not work very well; it does not search in answers at all.  However, coming in laterally via Google search and Google Alerts expands your possibilities exponentially [1].

[1] http://www.google.com/search?q=site:quora.com+Marcus-Endicott
= = =

Language Detection: What is a good way to strip a text of language-independent punctuation, like !, ?, and emoticons before trying for language detection?

This is usually referred to as Text Normalization [1].  See Vineet Yadav's answer to my Quora question: How would you make an API that converts any tweet into a proper English sentence?  In fact, I use Yahoo! Pipes Regex module for doing this.

[1] http://en.wikipedia.org/wiki/Text_normalization
= = =

Robots: I'd like to build a talking animatronic Beethoven bust for my music room, and don't even know where to begin, how to get started?

You can start by checking out my Meta Guide webpages, "100 Best Animatronic Head Videos" [1], "100 Best Robot Head Videos" [2], "100 Best Hanson Robotics Videos" [3], and "Talking Heads (Draft)" [4].

There are two aspects to this project, internal and external.  In fact, I usually recommend people create first a virtual talking head, for instance in a virtual world, such as Second Life.  Second Life is really a great testbed for protoyping robots.  A virtual prototype basically covers all of your internal workings.  Later, you can apply the virtual prototype to an external or robotic talking head.  For this, you can go with a generic robotic head, and simply dress it up after the fact as the character of your choice.

So-called rear projection talking heads seem to be the current favorite; however, there are also "virtual manequins" [5] available, though perhaps something of a compromise.  To do this well is actually quite a big and potentially costly project; however, I believe that you can achieve satisfaction with just the virtual prototype phase, and get a much clearer picture of what you want and how to get there.

[1] http://www.meta-guide.com/home/talking-head/100-best-animatronic-head-videos

[2] http://www.meta-guide.com/home/talking-head/100-best-robot-head-videos

[3] http://www.meta-guide.com/home/about/best-of-the-best-videos/100-best-hanson-robotics-videos

[4] http://www.meta-guide.com/home/talking-head

[5] http://www.meta-guide.com/home/about/best-of-the-best-videos/best-virtual-mannequin-videos
= = =

Artificial Intelligence: Can computers make questions?

"Question Generation" is now a scientific discipline in its own right [1].  I have previously made a number of quick and dirty Meta Guide webpages on "Question Generation" [2], "Question Generators" [3], and "Question Generator Module" [4].  However, there is a huge difference between a computer making questions and a computer "feeling the hunger to learn"; though, I can imagine a scenario in which a computer would even frantically pursue knowledge about something it did not already know about.

[1] http://www.questiongeneration.org

[2] http://www.meta-guide.com/home/bibliography/google-scholar/question-generation

[3] http://www.meta-guide.com/home/bibliography/google-scholar/question-generators

[4] http://www.meta-guide.com/home/bibliography/google-scholar/question-generator-module
= = =

The Future: Are robots, automation, AI, and Computers going to make us all unemployed?

Artificial intelligence will make "Republicans" obsolete; because, the only way forward in a labor force dominated by robots is Socialism [1].  Andrew McAfee @amcafee confirms .. robots will make Republicans obsolete .. Race Against the Machine [2].


[1] http://twitter.com/mendicot/status/199813134825955328

[2] http://twitter.com/mendicot/status/234323030241845248
= = =

What software stack does Siri use?

For some additional detail, see my answer to: Which topics on Quora are related to the technology behind Apple Siri?

= = =

Did IBM publish papers describing the algorithms used in IBM Watson?

Supposedly IBM Watson has 41 different subsystems (which I find analogous to Apple Siri's 35+ APIs).  I have not seen a complete list of these 41 subsystems; however, the IBM Journal of Research and Development recently published a special issue, "This is Watson" [1].  The electronic version with 17 papers at $31 each would run $527.  Despite a number of direct inquiries, I have been unable to locate any printed version.  :-(

[1] http://ieeexplore.ieee.org/xpl/tocresult.jsp?reload=true&isnumber=6177717
= = =

Q&A Websites: Perhaps you've posted or encountered an unanswered or poorly answered  question on Quora (or elsewhere) that you want to be answered well. Would you pay—real money, not credits—for a good answer?

Disintermediation... the real question is whether or not Quora is disintermediating consulting.  Solutions... solutions solve problems.  Gamification... could Quora credits be a version of neo-scrip [1]?  It would be nice if Quora credits could be donated to support some charity of one's choice, at least contributing towards a solution for someone.

I'm not at all convinced that you could buy the kind of really good answers one gets on Quora, as if serendipitously.  Nor am I convinced that unanswerable Quora questions could even be answered satisfactorily for money.

[1] http://en.wikipedia.org/wiki/Scrip
= = =

Social Media: What are the benefits of social bots indexing social chatter?
Social media monitoring [1] is most often used for opinion mining [2].

[1] http://en.wikipedia.org/wiki/Social_CRM

[2] http://en.wikipedia.org/wiki/Sentiment_analysis
= = =

Yahoo! Pipes: How do I create RSS feed out of images?

If these images can be viewed in a browser then they have a URL.  RSS requires elements or fields [1].  Yahoo! Pipes is *great* in that it allows you to monkey around freely with all the elements, below.  For instance, if you start with just a URL, or list of URLs (TXT/CSV), you can copy it to all the fields (pubDate optional), using the "Create RSS" module.  Then you can attack each field or element using the "Regex" module to add, subtract or replace basically anything you want.  I've found that if you run your Pipe output through Feedburner, it can correct certain inconsistencies, such as lack of proper pubDate.  I recommend starting with my Meta Guide webpage, "100 Best Yahoo Pipes Videos" [2].

- title
- description
- link
- guid
- pubDate

[1] http://en.wikipedia.org/wiki/RSS#Example

[2] http://www.meta-guide.com/home/about/best-of-the-best-videos/100-best-yahoo-pipes-videos
= = =

Q&A Websites: What are the key challenges in designing/developing an effective question answering system?

You first have to define what kind of question answering system you are on about.  Many people refer to the software behind QA boards and forums, such as Quora, as a "question answering system"....   QA boards and forums are generally a form of crowd-sourced QA, which falls under the rubric of collective intelligence.  Then there are fully automated question answering systems, such as Siri, generally referred to as artificial intelligence.  Often question answer pairs may be originally crowd-sourced, and then be subsequently applied to AI.

Increasingly, at the enterprise level, CRM support systems may start with an in-house "knowledgebase" that has previously been "crowd-sourced", for instance something like a "FAQ" - think question answer pairs.  Hybrid AI may then be applied to such an in-house knowledgebase, in the form of a question answering system; but on top of this, live support agents (people) may be integrated *transparently* to the end user to handle escalations from the AI - think "mechanical turk".  Thus, question answer "corpora" represent a kind of crowd-sourced "training" for the AI.

Make no mistake, Quora co-founder Adam D'Angelo is widely reported to have invested in Vicarious Systems - to build "software that thinks and learns like a human"....  The center of gravity in QA [1] research has been the on-going series of workshops known as TREC (Text REtrieval Conference) [2], sponsored by US government agencies; for instance, IBM Watson [3] stems largely from IBM's involvement in TREC.

[1] http://en.wikipedia.org/wiki/Question_answering

[2] http://en.wikipedia.org/wiki/Text_Retrieval_Conference

[3] http://www.mendicott.com/2011/01/how-many-playstations-make-watson.html
= = =

Starting a Business: How would one establish a home automation business in Venezuela?

Approach major home automation suppliers in the US about becoming their exclusive distributor for Venezuela.  See my Meta Guide webpage, "Dialog Systems in Smart Home Automation & Domotics (Domestic Robotics)" [1].

[1] http://www.meta-guide.com/home/ai-engine/home-automation

If you're going to identify major suppliers in home automation, you need to start somewhere.  I bet there are plenty of home security contractors in Venezuela; you could start by supplying them better voice control automation.... 
= = =

Robotics: Where do you see robotics in 5, 10 and 20 years?

Talking about the "Lego-ization" of cloud robotics ... see my recent videos on "Open Chatbot Standards for a Modular Chatbot Framework" [1].  

And see also particularly Justin Kiggins answer to: Neuroscience: Where can you visualize ideas as physical arrangements of neural networks? 

[1] http://www.meta-guide.com/home/open-chatbot-standards
= = =

Search Engine Optimization (SEO): How much web traffic does Linkedin send your site?

I noticed recently that I was getting more website traffic from LinkedIn by sending my tweets into LinkedIn than from Twitter itself, with roughly equal networks; that was right before Twitter cut LinkedIn off.  Since then I've used IFTTT to do the same thing, and actually prefer the tweets via IFTTT to the former, uglier Twitter branded ones.

Don't get me started on LinkedIn groups; see my question: LinkedIn Groups: Why is there no convenient way to follow LinkedIn groups?  This has got to be one of the biggest wasted opportunities on the Internet.  No feeds, no decent iPhone apps.... 50 groups worth of email digests is unusable.... :-(
= = =

Social CRM: Communicating with the new Social Customer, how should the ultimate Enterprise software look like?

See my recent answers to:

1) Marcus L Endicott's answer to Natural Language Processing: What are the state-of-the-art real world NLP applications on social media data? 

2) Marcus L Endicott's answer to Customer Service: Will avatars be a part of the future of customer service?

3) Marcus L Endicott's answer to What types of applications would you like to see for the Google AR Glasses?
= = =

Neuroscience: Where can you visualize ideas as physical arrangements of neural networks?

Where?  Why, just over the horizon....  You are on the right track with Connectomics though.  There is an inexorable top-down, bottom-up convergence underway between the low hanging fruit of the various neuroimaging technologies on the one hand, and brain simulation on the other.  

The best summary I've seen lately is the three Henry Markram videos in my recent answer to: Is it theoretically impossible for anyone to gain immortality through living in a machine?  I've also got more tools, projects and videos listed on my Meta Guide webpage, "Brain Simulation" [1].

[1] http://www.meta-guide.com/home/robopsychology/brain-simulation
= = =

Customer Service: Will avatars be a part of the future of customer service?

Often referred to as "holograms", they are in fact a rear projection mylar technology, a modern version of a classic stage trick called "Pepper's ghost" [1], and are otherwise known as "virtual mannequins"; for more detail, see my Meta Guide webpage, "Best Virtual Mannequin Videos" [2].  Generally thought to be one-way video, @Equicross claims their @AirportAvatar virtual mannequins are already "capable of answering spoken questions" [3].

Social CRM, including Mobile CRM, is undoubtedly heating up at the moment.  One just has to look back at how ATMs have transformed the banking industry to imagine the future full of synthetic people, or virtual humans, not to mention the coming augmented reality....

[1] http://en.wikipedia.org/wiki/Pepper's_ghost

[2] http://www.meta-guide.com/home/about/best-of-the-best-videos/best-virtual-mannequin-videos

[3] AVAARN Ticketing YouTube

= = =

IFTTT: Are there other companies offering or working on services like ifttt (ifthisthenthat)?

I am now categorizing consumer oriented services such as IFTTT, Yahoo! Pipes, Tarpipe, etc. together with iPaaS (integration PaaS), and would be happy to hear any arguments to the contrary.  In this regard, I can recommend my recent Meta Guide survey of iPaaS [1] providers.

[1] http://www.meta-guide.com/home/ipaas-integration-paas
= = =

Cloud Computing: What are the differences between cloud service providers? What must a cloud provider do to differentiate himself?

Cloud means cloud.  I want everything in the cloud, and I want it visual.

- I do not like having to download desktop environments; for me, this is just not cloud. 
- There is the command-line cloud, and the visual cloud.  I mean if I'm gonna use command-line, why would I want cloud in the first place? 
- Cloud providers who do not list their pricing on the site; I mean why would I waste my time with them?  
- Cloud providers who do not provide a free trial; I mean, if I can't play around with it and make sure it works for me, why bother? 
- 140 character descriptions, if a company cannot describe itself succinctly, why should I bother? 
- Videos, I want to see how the darn thing works in action, simple screencasts are fine; I don't want to see advertising from people who couldn't program if their lives depended on it, and I certainly don't want to see management scoping and pontificating about market placement.
- Twitter, if a company is not savvy enough to be working Twitter, honestly I find it hard to take them seriously in this day and age.

See my recent Meta Guide survey of iPaaS (Integration PaaS) [1] providers.

[1] http://www.meta-guide.com/home/ipaas-integration-paas
= = =
Yahoo! Pipes: What is the best resource to teach Yahoo Pipes to a novice that is not published by Yahoo or is a video?

I suggest starting with my own Meta Guide webpage, "100 Best Yahoo Pipes Videos" [1], which is actually only 62x at the moment.  I also have articles, books and example pipes listed under "Best Twitter Bot Pipes" [2].

[1] http://www.meta-guide.com/home/about/best-of-the-best-videos/100-best-yahoo-pipes-videos

[2] http://www.meta-guide.com/home/twitterbots/best-twitter-bot-pipes

= = =

Twitter: Can you filter out retweets on a Twitter feed?

Your solution is Yahoo! Pipes.

= = =
Text Analytics: How could a text summarization application be endowed with "personalities"?

Don't put the cart before the horse.  Virtual personalities need text summarization much more than vice versa.  Chatbots, conversational agents, and virtual assistants need text summarization as a component or subsystem in order to "interpret" various findings.  

Machine learning, classification can tell the difference between male and female writing, as well as identify both Hemingway and Churchill styles; why couldn't that be reversed for "machine teaching" in order to generate those styles?

= = =

Wolfram|Alpha: How can Wolfram Alpha be improved?

It needs better natural language understanding and natural language expression.  I would like to simply input the URL of my website, have it learn everything on my site, then be able to cut and paste an HTML widget into my site, like a chatbox, that can answer natural language questions based on the knowledge contained in my site, blog, or Twitter....  ;^)

= = =

Web Scraping: What's the name of the startup founded by Stanford grads that scrapes blogs and news sites for their article content?

Diffbot

= = =

Natural Language Processing: Which topics on Quora are related to the technology behind Apple Siri?

According to ProgrammableWeb, Apple Siri uses more than 35 APIs [1].  For instance, Siri accounts for about a quarter of the queries fielded by Wolfram|Alpha [2].  Yelp is another of the APIs Siri uses [3].

Originally, Siri was built on SRI International CALO (Cognitive Assistant that Learns and Organizes) [4].  What is not so widely known is that SPARK (SRI Procedural Agent Realization Kit) [5] forms the heart of CALO's task execution.

It's fairly certain that Nuance Communications ASR (Automatic Speech Recognition) is behind Siri [6].  And in terms of Cloud Computing, apparently Apple's new solar-powered data center in North Carolina "is handling at least part of the workload for Apple’s Siri voice command system" [7].

[1] http://blog.programmableweb.com/2011/10/04/are-there-open-apis-behind-apples-new-voice-commands/

[2] http://www.nytimes.com/2012/02/07/technology/wolfram-a-search-engine-finds-answers-within-itself.html

[3] http://officialblog.yelp.com/2012/03/your-iphone-personal-assistant-now-with-more-yelp.html

[4] http://en.wikipedia.org/wiki/CALO

[5] http://www.ai.sri.com/~spark/

[6] http://techcrunch.com/2011/05/09/apple-nuance-data-center-deal/

[7] http://www.macobserver.com/tmo/article/apples_north_carolina_data_center_adding_solar_power/

= = =

Text Analytics: Are there any online Bayesian SaaS text classifiers apart from uClassify.com?

In addition to uClassify.com, I've found BigML API [1] and Prior Knowledge Veritable API [2].  And, don't forget the Google Prediction API [3].  I don't know if they are all "Bayesian" per se, and would like to learn more about the pros and cons of each one myself.

[1] http://bigml.com/developers

[2] http://dev.priorknowledge.com

[3] http://developers.google.com/prediction

= = =

iPhone Applications: What are the most popular mobile app development applications, and what are their strengths/weaknesses?

I just did a survey of "Best App Building Tools" [1].  I believe it to be perhaps the most comprehensive of its kind.  There seem to be a very wide variety of options available.  It also seems that people are focussing more on HTML5 than native apps.  And, there is the school of thought that APIs come before apps....

[1] http://www.meta-guide.com/home/about/app-building-tools

= = =

RSS Readers: What's the best image-focused RSS reader app for iOS?

It's got to be Flipboard ..

= = =

Natural Language Processing: What is the best way to do simple Language Processing?

It's true that natural language is likely even more complex than we realize.  However, I've been looking at this field for a long time now, and believe that AIML (Artificial Intelligence Markup Language) [1] is a good, relatively simple place to begin working with sentence patterns, known as "pattern matching" [2].  And, I can recommend my own Meta Guide webpages "AIML Resources" [3] and "Best AIML Videos" [4] as starting points.

[1] http://en.wikipedia.org/wiki/AIML

[2] http://en.wikipedia.org/wiki/Pattern_matching

[3] http://www.meta-guide.com/home/ai-engine/aiml-resources

[4] http://www.meta-guide.com/home/ai-engine/best-aiml-videos

= = =

Natural Language Processing: What are the state-of-the-art real world NLP applications on social media data?

This cuts two ways.  There is a whole universe of Sentiment Analysis applications for social media; however, there are also the natural language understanding (NLU) [1] applications, such as conversational agents and social bots, increasingly used  for instance in social CRM.  Examples of social bots include variations on "autoblogging" platforms, such as:

a) http://rep.licants.org

b) http://www.weavrs.com

c) http://github.com/wilson428/Robottke


[1] http://en.wikipedia.org/wiki/Natural_language_understanding

= = =

Natural Language Processing: What tools can take a natural language query and convert it into a set of filters?

This is the proverbial "$64,000 Question".  I don't know everything about this field, and will be happy to learn more myself.  I have been investigating natural language tools for the past five years, and can recommend my recent videos on "Open Chatbot Standards for a Modular Chatbot Framework" [1].

There are tags, and there are tags.  Not all tag sets are created equal.  So, to a large degree, it depends on *how* your material is tagged.  Beyond that, it depends on *what* you have tagged.  Normally one would tag a natural language corpus, for instance a book or screenplay, for natural language interpretation.  Tagged "data", for instance below the sentence level, is another beast, and would need to be tackled statistically, or probabilistically.

This gets us into the nitty gritty.  The tool that is required is usually referred to as a natural language *interpreter*.  Let's just take AIML and Alicebot as the prototypical example.  AIML is a language, in other words a set of tags, specific to the various Alicebot interpreters.  There are "Alicebot" interpreters available now for most common programming languages.  AIML and the Alicebot interpreters are so-called "pattern matching" systems.  In terms of "filtering", this is basically how pattern matching interpreters work, by "filtering" on the tagged patterns, simply performing a kind of search.

Most chatbot hobbyists use pattern matching systems, of which there are many examples, often with their own "language" or tag sets.  Many of the Loebner Prize Turing test crew also use pattern matching, or hybrid statistical pattern systems, and often are involved developing their own language (tag set) and interpreter.  So-called "real" AI researchers tend to pooh-pooh even the Loebner Prize developers for using pattern matching techniques.  

The primary alternatives to pattern matching systems and their "proprietary" tag sets are statistical interpreters, think n-gram and "latent semantic".  There are not a lot of good examples of turnkey statistical natural language interpreters in common use.  Theoretically, statistical interpreters do not depend on tag sets.  However, there are hybrid systems which process tags, or patterns, statistically.  Patterns per se are just one kind of tag set, increasingly there are also "semantic" tag sets available.

Most grammar-based NLP tools seem to be more involved on the tagging side than the interpretation side.  As far as I know, there is no good, off the shelf natural language interpreter available for semantic web tagging, such as RDF.  There have been a number of attempts, and some claims; but, I've seen nothing concrete yet.  This is common in the AI world, where the higher one rises, the more nebulous they become, until literally disappearing from reality altogether....  You may wish to look at my list of, currently 85x, "Theses in AI & NLP" (from the past 10 years) [2].

[1] http://www.meta-guide.com/home/open-chatbot-standards

[2] http://www.meta-guide.com/home/bibliography/theses

= = =

Natural Language Processing: Is there a natural language processor in existence that can actually extract semantic content  from real human conversation?

Keith, I think you know the answer to this as well as anyone.  But, I'll take a stab at it.  First of all, it must be assumed that there is an entire layer of technologies, let us say, outside the public domain.  Secondly, as can be seen from Siri and Watson the entire ecosystem is tremendously complex, from hardware to software and even to crowdsourcing.  Thirdly, let's take the Turing perspective, and just say that I do know some machines more semantically competent than certain people.  

There is still a long way to go in terms of hardware improvements, not to mention shifting the installed base.  And, software to differentiate and follow different speakers in crowded rooms, for instance, is not yet at hand.  Further, what in fact is semantics?  When you drill down into grammar far enough it becomes extraordinarily vague as to what it actually represents.  My particular focus at the moment is on modeling the mechanism of converting words into images, and images into words in the human brain.  It is probably mathematical, and may prove to resemble something like fractals.  I can refer you to my draft webpages on Computational Metaphorics [1] in Computational Dreaming [2]. 

[1] http://www.meta-guide.com/home/bibliography/google-scholar/computational-metaphorics

[2] http://www.meta-guide.com/home/bibliography/google-scholar/computational-dreaming

= = =

Philosophy: Who am I? Am I my brain?

This one is a slippery slope; but, I'll give it a go....  The so-called "Mind–body problem" [1] is a classical philosophical issue, and of course relates to the eternal issue of "Soul" [2].  I believe that it is widely accepted today that mind is an "emergent property" [3] of brain; whereas, self as identity is very likely illusory [4], albeit very useful (like one's hand for instance)....  So, I guess in this sense, you are not you; therefore, your brain is also not you....  Who are you?  At the end of the day, you are everything....  ;^)

[1] http://en.wikipedia.org/wiki/Mind–body_problem

[2] http://en.wikipedia.org/wiki/Soul

[3] http://en.wikipedia.org/wiki/Emergent_properties

[4] http://en.wikipedia.org/wiki/Self

= = =

Speech to Text: What open-sourced and accurate speech-to-text engines and APIs currently exist?

Instead of speech-to-text (STT), Wikipedia has "Speech recognition" [1], otherwise known as automatic speech recognition (ASR).  Wikipedia also has a higher level "Category:Speech recognition" [2]; under which, you can find a "List of speech recognition software" [3].  In terms of open source, Wikipedia includes an entry for "Speech recognition in Linux" [4].

Of course, Nuance is the industry leader.  AT&T is also promoting their "Watson Voice Recognition Technology & Speech API" [5].  Red Shift Company [6] offers "RASR Speech Recognizer".  Koemei API [7] offers speech to text for video transcription.  Google also has their undocumented HTML5 Chrome speech API.  And in Windows, there is the Microsoft Speech API [8].

In terms of open source, CMUSphinx and their PocketSphinx [9] are probably most popular.  There is also the iOS version of PocketSphinx, called OpenEars [10].  There is an open source JavaScript SpeechAPI [11], similar to the MIT WAMI (Web-Accessible Multimodal Applications) toolkit [12].  Julius [13] is an open source example of "large vocabulary continuous speech recognition" (LVCSR).

[1] http://en.wikipedia.org/wiki/Speech_recognition

[2] http://en.wikipedia.org/wiki/Category:Speech_recognition

[3] http://en.wikipedia.org/wiki/List_of_speech_recognition_software

[4] http://en.wikipedia.org/wiki/Speech_recognition_in_Linux

[5] http://www.research.att.com/projects/WATSON

[6] http://www.redshiftcompany.com

[7] http://koemei.com/api

[8] http://en.wikipedia.org/wiki/Microsoft_Speech_API

[9] http://cmusphinx.sourceforge.net/?s=pocketsphinx

[10] http://www.politepix.com/openears

[11] http://speechapi.com

[12] http://wami.csail.mit.edu

[13] http://en.wikipedia.org/wiki/Julius_(software)

= = =

Cloud Computing: Is there a cloud framework?

As far as I'm concerned, people cannot agree on what a framework really is, much less what cloud is.  I believe the Wikipedia "Category:Cloud platforms" [1] conforms most closely to Infrastructure-as-a-Service (IaaS) and/or Platform-as-a-Service (PaaS) .  Myself I am most Interested in cloud middleware, but have only found a few visual IDEs; see my recent survey of "Artificial Intelligence Middleware" [2].  You may also be interested in my survey of "Application Programming Interfaces" [3].

[1] http://en.wikipedia.org/wiki/Category:Cloud_platforms

[2] http://www.meta-guide.com/home/artificial-intelligence-middleware

[3] http://www.meta-guide.com/home/application-programming-interfaces

= = =

Yahoo! Pipes: Do you know about good alternative to Yahoo Pipes?

I feel like this is a very important question, but do not have a clear answer yet.  I'm worried that the deteriorating condition of Yahoo Inc may not bode well for the future of Yahoo Pipes....  I believe that DERI Pipes is not currently operational.  Of course, the *idea* of DERI Pipes is fantastic, just like the *idea* of Yahoo Pipes is great....  

I've been struggling some days with Superfeedr Superpipes [1], and haven't succeeded in getting it to work properly.  I recently found Cordys Process Factory, and made a webpage aggregating videos about it [2]; unfortunately so far, the various critiques about Cordys on the web seem to be holding true.  I found Cordys by checking Wikipedia Category:Mashup (web application hybrid) [3].  I'm not really happy with the "mashup platform" designation; I'm really more interested in cloud based middleware, or visual IDE.  See also my recent replies to What are the best characteristics of a middleware platform? & RSS: Is there a viable replacement for Feedburner for bulk feed management? .

[1] http://github.com/superfeedr/superpipes

[2] http://www.meta-guide.com/home/twitterbots/best-cordys-process-factory-videos

[3] http://en.wikipedia.org/wiki/Category:Mashup_(web_application_hybrid)

= = =

What are the best characteristics of a middleware platform?

Yahoo! Pipes is my idea of a perfect cloud based middleware.  The problem is that the performance is not enterprise grade.  I would be happy to pay a reasonable amount for a premium Pipes service with better performance.  I'm not clear on the distinction between cloud based middleware and visual IDE.  Yahoo! Pipes has both web service and YQL modules providing for a high level of customizations.  It also has regex modules for infinite variety of processing.  Further, Pipes outputs in numerous formats, including json, csv, html, etc.  I'm doing a wide variety of things with Pipes, way beyond just processing feeds.  I can refer you to my recent survey of "Artificial Intelligence Middleware" [1].

[1] http://www.meta-guide.com/home/artificial-intelligence-middleware

= = =

RSS: Is there a viable replacement for Feedburner for bulk feed management?

This is a good question; I'd like to know the answer.  I use Yahoo! Pipes a lot.  I love the idea of Yahoo Pipes; but, lots of things on the Internet do not necessarily always perform as advertised.  I would be happy to pay a reasonable amount for a premium Pipes service with better performance.  I actually wish Google Reader would merge with Feedburner, just buy Yahoo Pipes, and mash them all together.  ;^)

There is Superfeedr; but honestly, I haven't been able to figure out how to use it yet.  The main problem I'm having is understanding what it does and how it works from a theoretical perspective.  I've been busting my nut for a few days trying to get Superpipes [1] to work properly, but without success yet.

[1] http://github.com/superfeedr/superpipes

= = =

Need to build a summarizer! That's okay. But what are the current sub problems that I could address in Summarization? Or what would you like a summarizer do for you? Thinking of a kick ass problem statement.

Think web feeds, think web APIs....  I want a sentence recognizer, and sentence extractor, based on grammar, not punctuation.  I want to extract complete and proper English sentences from web feeds.  I also want to filter web feeds based on grammatical correctness, correct=go and incorrect=stop.  Beyond this I want text normalization for web feeds, for instance correcting SMS language or translating Twitter rubbish into full English sentences.  And, I want all this in convenient web APIs.  ;^)

= = =

Are there any RSS feed autodiscovery tools that output feed URL when given a site URL?

I believe you could do this using the Yahoo! Pipes "Feed Auto-Discovery Module" [1].  Use the Fetch CSV Module then try putting the Feed Auto-Discovery into a Loop module.  Pipes will output to CSV using "_render=csv".

[1] http://pipes.yahoo.com/pipes/docs?doc=sources#FeedAutoDiscovery

= = =

Natural Language Processing: What is the current status of systems which can answer questions framed from a given text article?

Question Generation (QG) [1] and Question Answering (QA) [2] are two different things.  In terms of question answering systems (QA systems), I suggest looking at the Ephyra Question Answering System (OpenEphyra) [3].

As part of my Meta-Guide.com project, I'm currently running an experiment on Twitter, parsing travel questions off the Twitter firehose and running them into two different conversational agents (chatbots) at @VagaBot [4] and @TwavelAdvisor [5].  Over the past year I have accumulated more than 36,000 unique travel questions.

[1] http://www.questiongeneration.org

[2] http://en.wikipedia.org/wiki/Question_answering

[3] http://www.ephyra.info

[4] http://twitter.com/vagabot

[5] http://twitter.com/twaveladvisor

= = =

Can a computer that is designed to fake self-awareness be as good as a computer that is truly self-aware?

Think "smoke and mirrors"....  "Mind games" are a HUGE part of any Turing class bot; in fact, botmasters regularly program intentional mistakes to accurately simulate human fallibility in order to fool human Turing test judges.  Natural language dialog is hugely psychological.  My father often said that it doesn't matter what you "say", only what others "hear".  

Secondly, self-awareness in a machine will very likely be different from self-consciousness in a human; so, machines may very well become self-aware, but not humanly self-conscious.  Self-awareness in a machine could very well manifest as reflective environmental consciousness.  A machine may become more aware of its surroundings than a human.  A machine may know everything about itself, and in a practical sense more than any human knows about itself.  Then the machine could calculate every relation and every possibility between itself and the environment, for instance like a prize winning chess computer.  The result would be that the machine was much more self-aware than the person standing next to it.

But this heightened self-awareness would be superior to that of the human; therefore, it would not fool any human.  In fact, the only way to fool a human would be to "dumb down" the machine, ergo smoke and mirrors mind games; so in the end, a machine with human level consciousness would be inferior to the truly self-aware machine.  ;^)

= = =

Conversational Agents (Chatbots): I want to create a custom chat bot for use with smart home software, but not from scratch. What's the best method to do this?

In lieu of answering this question here, I have made a new Meta-Guide.com webpage on "Dialog Systems in Smart Home Automation & Domotics (Domestic Robotics)" [1].  This webpage was created by extracting all relevant references from my four years of tweets as well as from my Meta Guide website itself.  

Although not open source, I imagine that the new Microsoft HomeOS [2] could be used with the Kinect for Windows [3], which has speech capabilities.  (See also Microsoft Robotics Developer Studio microsoft.com/robotics ..)

I can add that Steve Prior [4] is a good person to ask about dialog systems in smart homes in general, and about his "Geekster Smarthome" [5] in particular.

[1] http://www.meta-guide.com/home/ai-engine/home-automation

[2] http://research.microsoft.com/en-us/projects/homeos

[3] http://www.kinectforwindows.org

[4] http://twitter.com/sprior

[5] http://smarthome.geekster.com

= = =

What are some of the best chatterbots?

Cleverbot.com [1] is arguably the most popular.  The newly popular SimiSimi mobile app is perhaps a close second, particularly in Asia.  Wikipedia contains both a Category:Chatterbots [3] and a List of chatterbots [4].  Chatbots.org also maintains a relatively up to date Directory [5]. 

As for what "best" really means, the sky is the limit....  Guile3d.com Denise [6] is probably the best full featured desktop virtual assistant at the moment.  There are a number of learning bots available, such as Zabaware - Ultra Hal Assistant [7].  Ultra Hal is also apparently learning continuously in the cloud from Twitter, and can be interacted with on Twitter @UltraHal [8].  @twthal [9] is another interactive learning chatbot on Twitter by Ai Research (a-i.com).

[1] http://en.wikipedia.org/wiki/Cleverbot

[2] http://en.wikipedia.org/wiki/SimSimi

[3] http://en.wikipedia.org/wiki/Category:Chatterbots

[4] http://en.wikipedia.org/wiki/List_of_chatterbots

[5] http://www.chatbots.org

[6] http://guile3d.com/en

[7] http://www.zabaware.com/assistant

[8] http://twitter.com/ultrahal

[9] http://twitter.com/twthal

= = =

Artificial Intelligence: How would a society benefit from an AI that passes the Turing test?

It is my belief that the Turing test is a "red herring", which is not to say that society might not benefit in any number of ways from an AI that passes the Turing test.  I also recognize that the pursuit of the Turing test stimulates development, also in any number of ways.  

The Turing test is a red herring; because, it is a distraction from the more urgent necessities of developing AI that are not only helpful to people, without fooling them in any way, but from the even greater task of developing AIs that will surpass humans in any number of specific domains or verticals, within this century.  For instance, a computer system with access to all of recorded history could very likely predict outcomes better than mere mortals, in other words know what people will most likely do before the people themselves know.

In general, the benefits of artificial intelligence are clear to me, especially when coupled with the coming "Internet of things", sensors etc.  Arbitrating infinitely complex environmental management tasks, such as effectively addressing global warming and climate change, obviously requires intelligence superior to that of humanity....

= = =

What types of applications would you like to see for the Google AR Glasses?

Think #SecondLife ... I would like to see avatars move back and forth between virtual worlds and augmented reality.  I want voice-interactive NPC chatbots in augmented reality.  I would like to see a *platform* where anyone can launch their own embodied #Siri or #Watson; ideally, this would be some kind of pluggable middleware, where people could mix and match any cloud intelligence, or API, with a myriad of customizable 3D animations ala Second Life.  ;^)

= = =

Linguistics: Is it possible to explain metaphors using discourse analysis frameworks, or only with cognitive linguistic frameworks?

If by "discourse analysis frameworks" [1] you are referring to "Metaphor Identification Procedure", and by "cognitive linguistic frameworks" [2] to "Image Schema", then I suggest the answer lies in between....  ;^)  

For me the real question is... what are the mechanics of converting words into images, and images into words in the human brain?  I tend to think of this mechanism in terms of a "clutch", or transducer of some kind, and can refer you to my draft webpages on the role of Computational Metaphorics [3] in Computational Dreaming [4].

[1] http://en.wikipedia.org/wiki/Category:Discourse_analysis

[2] http://en.wikipedia.org/wiki/Category:Cognitive_linguistics

[3] http://www.meta-guide.com/home/bibliography/google-scholar/computational-metaphorics

[4] http://www.meta-guide.com/home/bibliography/google-scholar/computational-dreaming

= = =

Natural Language Processing: What are the most popular frameworks used for creating natural language user interfaces?

The term "natural language", like the term "artificial intelligence", is really non-specific, covering a wide range of sub-fields.  Unfortunately, frameworks and interfaces in this context are equally vague, and ill-defined.  This doesn't reflect on the questioner, but is simply inherent in the fields at this point.

There are really no standards for "natural language interfaces" yet.  And, as far as I'm concerned, people can't even agree on what's a proper framework.  However, Wikipedia does include a "List of natural language processing toolkits" [1].  Of these, NLTK (Python) is probably the most popular.  Stanford NLP is also highly spoken of.  GATE and Lingpipe are used quite a bit too.  Myself, I'm leaning toward the Web APIs, such as AlchemyAPI.

However, an NLP toolkit does not an interface make.  For instance, many of these toolkits are used within the Eclipse [2] integrated development environment (IDE), which might be more properly viewed as a framework for NLP libraries.  Further, speech is a different beast from NLP, and the tools for speech interfaces are very often not the same as for NLP per se.  And, the real challenge at the moment is integrating Semantic Web technologies effectively with natural language interfaces.

Myself, I'm extending these concepts to cloud robotics, and am in search of proper AI middleware in the cloud for integration purposes; but, the closest I've come so far are cloud-based IDEs (aka Web IDEs), for example Heroku.  I have yet to find a real, cloud-based Visual IDE.  You can find more about my own journey into these realms on my Meta Guide project website [3].

[1] http://en.wikipedia.org/wiki/List_of_natural_language_processing_toolkits

[2] http://en.wikipedia.org/wiki/Eclipse_(software)

[3] http://www.meta-guide.com

= = =

Artificial Intelligence: Is IBM's Watson an Expert System?

Expert system implies to me rule-based AI.  When I was bandying around the conventional wisdom that IBM Watson represented the triumph of the era of machine learning over the past epoch of rule-based AI, and on Quora no less, IBM Watson researcher, and Quora member, Bill Murdock [1] soundly rejoined that such was not the case [2]....  ;^)

[1] James William Murdock IV

[2] How does IBM's Watson work?

BTW, David Ferrucci let slip recently that the IBM Systems Journal will be preparing a special issue dedicated to IBM Watson.

= = =

What are the repercussions of Twitter selling our tweets?

Personally, I feel some anger and resentment toward Twitter.  I believe that Twitter is more of an infrastructure than an online service.  I really think Twitter is going down the wrong road toward Facebook-ization.  Trying to improve the elegant simplicity of Twitter seems exactly like Windows-bloat-ification, in other words making it infinitely more complicated by trying to simplify it.  

Further, by instituting their own URL shortener, and thereby screwing up URLs as posted by users, Twitter is actually engaged in a massive trip to control URLs.  For example, if Twitter should go out of business at any time in the future, then every link ever posted to Twitter could be lost, not cool.  

As for the historical corpus of Twitter, IMHO, it is verging on criminal to deny the public access to their own archive.  Why is this important?  Because, almost everything about almost everything has already been said.  In other words, almost all knowledge is already contained in the historical corpus of Twitter, and throttling access to that knowledge is unconscionable!

= = =

How does the chatbot jabberwacky work?

I've previously answered a similar Quora question, "How does Cleverbot work?" [1], and have since blogged "My Cleverbot Tweet-FAQ" [2].  Both Jabberwacky and Cleverbot were created by Rollo Carpenter [3].  Jabberwacky [4] was created first (1997) by his Icogno.com and Cleverbot [5] later (2008) by Existor.com .  Presumably Rollo is on Twitter at Existor [6].  Notably, Rollo won the Loebner Prize in 2005 and 2006 with Jabberwacky / Icogno technology.  Jabberwacky / Icogno technology is now behind both LifeNaut.com and LiveEnglish.ru .

Supposedly, Cleverbot is a variant of Jabberwacky, but more fuzzy and with deeper context.  As for how Jabberwacky learns, Antonella De Angeli [7] and Sheryl Brahnam [8] published a 2008 academic paper, "I hate you! Disinhibition with virtual partners" [9], essentially an extensive analysis of the Jabberwacky knowledgebase.  They state that "it learns by association, storing replies to inputs in a database".  For instance, Cleverbot uses 4d.com (4D v11 SQL).

I have posted a webpage, an informal survey of the past 10 years of academic work mentioning Jabberwacky [10].  Jabberwacky is generally considered to use an AI technique called "contextual pattern matching".  I have not found evidence of any specific machine learning algorithms used in Jabberwacky; however, Jayen Ashar completed a 2010 PhD thesis entitled, "Online Learning in Conversational Agents" [11], essentially a literature review, which provides an overview of the field.

[1] How does Cleverbot work?

[2] http://www.mendicott.com/2011/07/my-cleverbot-tweet-faq.html

[3] http://en.wikipedia.org/wiki/Rollo_Carpenter

[4] http://en.wikipedia.org/wiki/Jabberwacky

[5] http://en.wikipedia.org/wiki/Cleverbot

[6] http://twitter.com/existor

[7] http://www.informatik.uni-trier.de/~ley/db/indices/a-tree/a/Angeli:Antonella_De.html

[8] http://www.informatik.uni-trier.de/~ley/db/indices/a-tree/b/Brahnam:Sheryl.html

[9] http://dac.escet.urjc.es/rvmaster/rvmaster/asignaturas/articulo.pdf

[10] http://www.meta-guide.com/home/bibliography/google-scholar/jabberwacky

[11] http://jayen.web.cse.unsw.edu.au/phd/litreview.pdf

= = =

What are the most beautiful word cloud designs?

Tagul.com claims to make "gorgeous tag clouds" .. ;^)
= = =

 
The real strength of Siri lies in its middleware infrastructure, that ties together more than 35 APIs. I've got a wealth of resources listed under my Meta-Guide.com project website. And, I've put together a comprehensive survey of "Artificial Intelligence Middleware" [1], as well as a webpage listing relevant "Application Programming Interfaces" [2].  

In fact, Siri is based on a government sponsored (SRI) project called PAL (Personalized Assistant that Learns), and *some* of the PAL Framework [3] components are open source. (Actually, CALO, the Cognitive Agent that Learns and Organizes, was originally developed as part of the PAL project.)




Comment • Jan 14, 2012

= = =

 
Firstly, I recommend checking my "Proceedings" page on Twylah [1].


Secondly, you might check the following:

Declarative Agent Languages and Technologies

Intelligent Virtual Agents

Interactive Digital Storytelling

International Conference on Virtual Storytelling

Web Reasoning and Rule Systems

Comment • Dec 20, 2011

= = =

 
"Computational Dreaming" is directly related to "Metaphor Analytics", which necessarily involve "imaging" and pattern recognition.

Comment • Dec 16, 2011

= = =

 
There are plenty of data visualization applications to see the spread of information on Twitter over time. I simply monitor Topsy [1] to see who says what, and when, about any given URL.

 
Your answer may need to be a more direct response. (more)

1 Comment • Dec 16, 2011

= = =

 
My Pipes seem to be working correctly at the moment. Pipes was down for a few days, but came back up this morning. There are lots of modules in Pipes, and any one of them can develop issues. I lost some vital functionality in the "Fetch Data" module after the recent upgrade to V2.

1 Comment • Dec 15, 2011

= = =


I suggest looking at OpenEphyra [1], a contender in the Text REtrieval Conference (TREC) [2] series, and precursor to IBM Watson.



Comment • Dec 12, 2011

= = =

 
I suggest reading Twitter's own "Automation Rules and Best Practices" [1]. I am currently building travel and tourism Twitter bots for every country on Earth [2]. I've also got an inventory of non-travel Twitter bots I've made [3]. 

At this time, Twitter uses automation to "suspend" abusive Twitter bots. There is an appeal process, which works for me maybe 50% of the time. In particular, Twitter does not like "@" sign abuse; I've taken to replacing them with "#" tags (after removing existing "#" tags). Also, Twitter does not like people ripping off tweets without attribution of some kind (such as, at least a link to the original).

Since twitterfeed.com is rate limited, I now prefer dlvr.it. (BTW, Twitterfeed was acquired by bitly.com not long ago.) I have also used tweethopper.com in the past. Actually, feedburner.com now includes the native ability to pass feeds into Twitter. And of course, there are others.




Comment • Dec 11, 2011

= = =

 
In addition to Alabot [1], Vimagino (vHelp.me) [2] and WEBees (Convogent) [3] work in the conversational agent (chatbot) space in India.




Comment • Dec 10, 2011

= = =

 
Basically, rules based AI has given way to machine learning. Unstructured data is "understood" by checking it in parallel against structured data, and statistically analyzing the evidence. 

Have a look at my January 2011 blog post, "How Many PlayStations Make A Watson?" [1]. I've also got a page of "Best IBM Watson Videos" [2].



Comment • Dec 4, 2011

= = =

 
Wavii appears to be a combination of automated "web content curation" [1] and so-called "text synthesis" [2], which would make it related to both Summify [3] and Automated Insights' [4] (formerly StatSheet) [5] STAT.US [6].







Comment • Dec 4, 2011

= = =

 
This has recently been addressed in: What are Paper.li competitors?

Comment • Dec 4, 2011

= = =


What immediately comes to my mind are the Boston Museum of Science "responsive virtual human museum guides", Ada and Grace [1], made at the University of Southern California Institute for Creative Technologies [2]. Certainly, such virtual human museum guides might also be implemented in augmented reality.

There is a new book available in this field, titled "Language Technology for Cultural Heritage" [3]. The AMICUS network [4] for "automated motif discovery in cultural heritage and scientific communication texts" is another available resource. And, a registry of technology-related museum projects is maintained online by MuseTech Central [5].






Comment • Nov 24, 2011

= =  =

 
"Human-like" AI is commonly referred to as AGI, or "Artificial General Intelligence" [1]. I believe that "sentiment analysis" is in fact more closely related to conversational agent (aka chatbot) NLP or AI than people seem to realize.

Keep in mind, there are two layers to "human-like" AI. One is the speech layer, or dialog system (using essentially the same technology as sentiment analysis, but in a different way). The second layer is the "cognitive", perhaps more like IBM Watson in that it's actually "understanding" concepts and their relations. So, the fusion of Apple Siri and IBM Watson would give some idea of a more "human-like" AI. Once Apple releases a Siri SDK, and IBM releases its Watson API, then people may begin to hack around and realize this potential. (Of course, the availability of robust AI middleware platforms would also be helpful.)

However, Seth Grimes is probably right that the "easiest" way forward would be with machine learning based on crowd-sourcing. Perhaps one day soon machines will be digesting and learning from Quora.


Comment • Nov 24, 2011

= = =

 
Gary, I have written extensively about the need for "Open Chatbot Standards" [1], and have frequently mentioned on Quora [2] the potential key role of XMPP for machine to machine natural language communication.



FYI, I have also recently aggregated links and videos about "Multi Bot Scenarios" [3], such as the viral "AI vs AI".


BTW, my personal feeling about the Turing test is that it is a "red herring".

Comment • Nov 24, 2011

= = =

 
I would like to expand on Kitano and O'Brien's answers. I am seeing 2 sectors here. One I'm calling "robojournalism", and the other "curation". Robojournalism may be "newspaper" or "magazine" formats. Robojournalism also includes "summarization"[1] or "text synthesis"[2], and there are topics for these on Quora. Curation may be automated or manual. 

Paper.li (@SmallRivers) and TweetedTimes.com (@TwtTimes) come under the newspaper format. And, Postano.com (@ilpostano), Twylah.com (@Twylah), and Scoop.it (@scoopit) qualify as magazine formats. There are differences here, some do your inside stream, some do outside stream, others follow hashtags or multiple streams, and Scoop.it is manual curation though magazine format.

So-called "summarizers" include Summify.com (@summify), Strawberryj.am (@strawberryapp), Knowabout.it (@knwbt), News.me (@newsdotme), and Topicmarks.com (@topicmarks).

Examples of more advanced "text synthesis" robojournalism include AutomatedInsights.com (@AInsights), formerly StatSheet, and NarrativeScience.com (@narrativesci).



Comment • Nov 21, 2011

= = =

 
It does not yet remember links. I would prefer to never receive the same link twice.

Apparently, links are triggered by 2x posts; so, it doesn't even check links for rank until it sees them twice. I would prefer to have all links checked on popularity outside my stream.

And then there is the inverse. Do I really want the most popular links (think TV)? Maybe the gold is really in the least popular ones?

1 Comment • Nov 20, 2011

= = =

 
I would say one is more "push", and the other more "pull". For example Apple Siri, if not a "recommender system" per se, certainly involves elements of recommendation systems; whereas, something like StumbleUpon.com would be more of a "discovery engine".

Comment • Nov 15, 2011

= = =

 
I've got a listing of (82x) conversational agent companies on Twitter at


Comment • Oct 18, 2011

= = =

 
I feel strongly that there will be a boom in XMPP for machine-to-machine natural language communication (AI vs. AI).

Extensible Messaging and Presence Protocol

Further, if there were any decent IM-Voice bridge (app) available, then not only would everyman have safer hands-free instant messaging, but would also be able to speak with their own AI using speech i/o.

1 Comment • Sep 12, 2011

= = =

 
It seems that "constructed languages" would lend themselves more readily to machine intelligence, because they are equivalently artificial, as opposed to natural language. In fact, Cleverbot.com is able to speak the constructed language "Toki Pona".

List of constructed languages - Wikipedia

On a tangential note, I am a great proponent of XMPP for machine-to-machine natural language communication (AI vs. AI).

Extensible Messaging and Presence Protocol

Comment • Sep 12, 2011

= = =

 
I just got back a few months ago from an extended trip, with my iPod Touch, to South America. The iTouch is great for travel because you don't need a plan for WiFi, and it does Skype just fine. 

Though there were a few choices, I finally settled on paying $9.99 for Wiki Offline by Avocado Hills [1]. This was by far the best app for the trip. I loved reading background and history, even for a surprising number of medium-sized places.


In contrast, Word Lens by Quest Visual proved to be a complete waste of money.

I found another unlikely app to be most handy, the free MapsWithMe - Travel Guide by Yury Melnichek [2]. This is great for two reasons. It had all the free offline local maps I needed for the whole trip, and contained all the offline Wikitravel I needed.


I really love the Web Reader - Text to Speech $1.99 app by Chris Chauvin [3]; because, I filled my Mendeley - Reference Manager app [4] with 1,000 research PDFs and listened to them to my hearts content. ;^)



Finally, I ended up buying the Justin.tv app [5] for $4.99 en route, just to get some free English language TV. Despite containing atrocious advertising for a paid app, it really surprised me what all you can watch on Justin.tv. ;^)


I never did locate a really good *offline* podcast app.

Comment • Sep 8, 2011

= = =

 
PersonalBrain Tour in 10 Minutes

Concept mapping versus topic maps and mind mapping

List of concept mapping and mind mapping software


I think I would vote for Mindjet.com MindManager (mainly because you can export from http://www.wikisummarizer.com into Mindjet).

Data visualization is a related field, which might give you some more ideas.


Data visualization software


(Disclaimer: I make artificial minds at http://meta-guide.com ..)

1+ Comments • Sep 6, 2011

= = =

 
I just ran across "Ping Identity Corporation" [1] the other day while researching the fate of the former Jabber.com (acquired by Cisco.com). It seems that Jabber, Inc founder Andre Durand moved on to found "The Cloud Identity Security Leader"(tm).


1+ Comments • Sep 6, 2011

= = =

 
Chat means one thing to me, XMPP. I would encourage you to look into Voxeo IMified at http://imified.com . I really don't think that people appreciate the significance of XMPP. XMPP is all about natural language communication. I feel strongly that in the future more machines will be using it than people. I believe that XMPP will be the primary means of machine to machine communication using natural language. I also believe that XMPP will be the way that people communicate with AIs; however, surprisingly, there is no consumer IM - Voice bridge currently available. That will be a revolutionary app, a simple bidirectional instant message to voice app, with universal XMPP plug and play on the backend - not only for hands-free IM, but also to actually speak with AIs.

Comment • Aug 29, 2011

= = =

 
Just a quick google shows about 750,000 results for site:pipes.yahoo.com


I think there could reasonably be around three quarters of a million pipes created. I would hazard to guess that there may be half that number of all time registered users, so about 375,000. I can say that I have 50,000 users on 50 pipes on Twitter, so 1,000 users per pipe. That would make some 750 million Yahoo! Pipes users total. (I will be happy to be proven or disproven on any of these numbers.)

Comment • Aug 25, 2011

= = =

 
Gary, as you probably know, there's been a lot of talk on Robitron and the Chatbots.org AIZone about Bruce Wilcox's ChatScript (versus AIML). There is also RiveScript by Noah Petherbridge, CML (Conversation Markup Language) by Chongguan Yang, and AIScript (PersonalityForge) by Benji Adams. The list goes on.

Current Loebner champion Wilcox has waxed eloquent about this in March 2011, "Beyond Façade: Pattern Matching for Natural Language Applications":


He revealed more details about his work in June 2011, "Suzette, the Most Human Computer":


As I've mentioned previously on Chatbots.org AIZone, its not so much the language as the interpreter that makes the most difference to the learning curve. (Though I do agree theoretically with the logic for the improvements of ChatScript over AIML.) Its in fact the (cloud) infrastructure of new services like Chatbot4u that make the most difference. It just feels like to me that the old days of hosting an interpreter on your own hardware are quickly passing away.

That said, I'm a great believer in MODULARITY, open plug and play frameworks. I believe, both Apple Siri and IBM Watson were made in this way. Currently, the biggest bottleneck is the voice-in/voice-out lipsync avatar. I am a big promoter of XMPP as the lingua franca for open, modular conversational agents. But, there is no good IM-Voice bridge available at this time (with or without avatar).

Comment • Aug 25, 2011

= = =

 
This is a hot topic now, and a lot is being written about it.

I work with speech recognition in conversational agents, and can tell you it's still not perfect. While this is amusing in chatbots, it's not quite ready for mission critical applications.

But voice search does not mean simply speech to text; voice search also implies text to speech. For instance, all of the Twitter readers I've tried are still imperfect and have a hard time with hashtags, URLs, and SMS language, which doesn't make for fun listening.

The big component that's still missing in almost all applications is summarization. Any voice search will need to at least partially abbreviate and interpret, if not tailor, results.

I imagine in the medium-term future people will cease to surf the web as they do today, and a layer of (voice) agents will use the web as we know it for their backend.

1+ Comments • Aug 23, 2011

= = =

 
I've just begun Alpha testing DataSift, and its not yet clear to me what I can do with DataSift that I'm not already doing with Yahoo! Pipes.

There is no clear feed-in / feed-out paradigm, since DataSift taps services internally:


To do anything in DataSift involves using its own Curated Stream Definition Language (CSDL):


For feed-out, DataSift more resembles Tarpipe in terms of pricing:



Comment • Aug 9, 2011

= = =

 
The holy grail of travel technology... the artificially intelligent, robotic travel agent, as smart or smarter than the average professional travel agent today. Smartphone apps such as Apple Siri and others are a step in that direction, but no cigar yet. In fact, I've been working on this for the past 13 years at Meta-Guide.com .

Comment • Aug 2, 2011

= = =

 
Siri - The Personal Assistant on your Phone

Monica iPhone Application Full Demo - Your Virtual Assistant

Meet Alfred, your personal robot!

Comment • Jul 31, 2011

= = =

 
In these new videos, Philadelphia lawyer Frank Taney http://twitter.com/scarylawyer of Buchanan Ingersoll & Rooney discusses legal and IP issues for botmasters at the recent Chatbots 3.1 Conference.

Chatbots 3.1 - Francis Taney - Legal and IP Issues for Botmasters (1/2)

Chatbots 3.1 - Francis Taney - Legal and IP Issues for Botmasters (2/2)

Comment • Jul 21, 2011

= = =

 
I can add three clues to this thread.

Firstly, I recently came across the 2009 book, "Introduction to Chinese Natural Language Processing"


by


Secondly, it seems Ben Goertzel has been working extensively in China with Australian Professor Hugo de Garis; so, I suspect either would know who's who and what's what.



Third, there appears to be significant recent work being done in China on metaphor analytics, which might explain the current interest in Washington.

2010: "The Chinese Noun Metaphors Knowledge Base and its Use in the Recognition of Metaphors"
by

2010: "Research on the Cognitive Comprehension Logic and Its Application in Understanding of Metaphor"
by

Comment • Jul 20, 2011

= = =

 
Stephen Wolfram wrote a great blog post in January 2011 comparing IBM Watson with Wolfram|Alpha side by side, entitled "Jeopardy, IBM, and Wolfram|Alpha".


Apparently Wolfram|Alpha uses Stephen Wolfram's own Mathematica software to supposedly compute symbolic representations.


I would welcome a more detailed explanation of just how this works myself! ;^)

I was able to find this Youku video of Stephen Wolfram's brother Conrad Wolfram explaining symbolic computation in Mathematica.


1 Comment • Jul 20, 2011

= = =

 
There is a Quora topic for this at 


Wikipedia is now referring to this as an "Automated online assistant"


The software may be referred to as a "Dialog system" (or Dialogue system)


See in particular "Toolkits and architectures"


Comment • Jul 18, 2011

= = =

 
"Artificial intelligence has the same relation to intelligence as artificial flowers have to flowers." David L Parnas

These are very recent, very cool videos that tell it like it is, and we ain't there yet.

Henry Markram: Simulating the Brain -- The Next Decisive Years [1/3]

Henry Markram: Simulating the Brain -- The Next Decisive Years [2/3]

Henry Markram: Simulating the Brain -- The Next Decisive Years [3/3]

1 Comment • Jul 17, 2011

= = =

 
I just ran across this 2010 book:

Automated Grammatical Error Detection for Language Learners

by 


Comment • Jul 17, 2011

= = =

 
This is an interesting concept; though, I've got a slightly different slant on it. There are two basic kinds of "artificial intelligence", one is conventional machine intelligence, and the other is "collective intelligence" along the "Mechanical Turk" model. (And of course, there is also the combination of the two.) It seems that you have identified a prime opportunity to implement this kind of solution, and not to forget the FUN (and learning)!

I will add to this that as things get worse in the world environmentally and socially, I expect people will retreat into virtual worlds, along the lines of games such as Second Life and World Of Warcraft. At the same time, these virtual worlds are predicted to become MUCH more realistic (probably hyper-realistic like HDTV). Therefore, the question may become how to employ people as "Mechanical Turks" in the virtual worlds.

Comment • Jul 15, 2011

= = =

 
This may be a somewhat tangential or ancillary answer, but is inline with my area of expertise, social agents for question answering.

According to "Agent Virtuel - LE BLOG" http://agent-virtuel.fr ... the main players in the "automated online assistant" industry in France are:

VirtuOz - Assistant Virtuel pour la Gestion du Service Client

Agent virtuel ASKOM: améliorer la relation client - guider, conseiller, humaniser

Do You Dream Up ? (aka createmyassistant.com)

Agent virtuel dialoguant, agent conversationnel - dialonics

Comment • Jul 13, 2011

= = =

 
AdaptiveAI is behind SmartAction, and Peter Voss is behind both. You can find his 2007 paper, "Essentials of General Intelligence: The Direct Path to Artificial General Intelligence", below:

a2i2 : Adaptive AI, Inc

Peter Voss | LinkedIn

Essentials of General Intelligence: The Direct Path to Artificial General Intelligence (2007)

Comment • Jul 8, 2011

= = =

 
Artingence - The Artificially Intelligent Call Center

SmartAction call center IVR, hosted speech IVR, IVR solutions

Comment • Jul 8, 2011

= = =

 

>>
The term reverse Turing test has no single clear definition, but has been used to describe various situations based on the Turing test in which the objective and/or one or more of the roles have been reversed between computers and humans.
<<


>>
(NLG) is the natural language processing task of generating natural language from a machine representation system such as a knowledge base or a logical form.
<<

Machine learning is getting pretty darn good, especially when coupled with brute force computing. I suspect that the time is not long off where machines will be able to identify humans more readily than humans will be able to identify machines.

Comment • Jul 5, 2011

= = =

 
There is no Wikipedia entry for "text synthesis".

However, there are Wikipedia entries for "Natural language generation" and "Automatic summarization":



There are recently answered questions on Quora in these topics:



On Wikipedia StatSheet is referred to as "automated publishing":


The only thing I could find on Wikipedia directly related to automated publishing was "Dynamic publishing":


There was a Quora question on "dynamic semantic publishing" which I've added to your "text synthesis" topic:


Comment • Jun 30, 2011

= = =

 

Cleverbot uses string metrics, a technique called "string similarity"..
See String metrics:  http://en.wikipedia.org/wiki/String_metric Cleverbot creator, Rollo Carpenter, discusses his work in a series of videos entitled "Learning Creating Phrasing" => 
http://www.youtube.com/playlist?list=PL5558B77F03AE612E

Comment • Jun 29, 2011

= = =

 
On the Wikipedia "List of natural language processing toolkits" http://en.wikipedia.org/wiki/List_of_natural_language_processing_toolkits .. AlchemyAPI Ruby SDK http://www.alchemyapi.com/tools/ seems to be the only one listed for Ruby ..

Addendum:

This Ruby-1.8 "Natural Language Processing (NLP) Software" by Masao Utiyama (last updated 2007) turned up after my original answer above:


Comment • Jun 26, 2011

= = =

 
Good question.  I know of at least two companies working in this area:
 
 
and
 
 
Offhand, I am not aware of turnkey products to accomplish your goal.  However, I recommend you look at the new Guile3D conversational agent (aka chatbot) Denise http://www.guile3d.com , basically taking advantage of Windows7 speech tools.
 
I have asked Richard Wallace, inventor of AIML, a number of times for a list of known voice interactive applications, but so far without success.  I have been told by Voxeo/Tropo that IVR Grammars are not up to this task, despite their announcement of partnering with http://artingence.com to do just this. 

What caught my attention while listening to the paper you cited (in iPhone Web Reader text to speech app) was the mention of confidence scoring, of the type used in IBM Watson, and the use of Ngrams, such as Microsoft Web N-gram Services or Google Ngram Viewer.

Comment • May 18, 2011

= = =

 
This question does not make sense.  Please clarify.  Do you mean Visual Basic or visually impaired??  Otherwise, Pipes is as visual as it gets; but, to do anything really creative you still need to do some programming, essentially creating your own webhooks.

1+ Comments • May 15, 2011

= = =

 
I certainly couldn't say who's the smartest NLP person in Poland; however, I can say that there are a number of chatbot specialists either based in Poland or originally from Poland. I hope you find the following five companies interesting.






Comment • Jan 30, 2011

= = =

 
Yes, its trivial to send Quora feeds into a chatbot or question answering API; however, chatbot replies tend to be "chatty" rather than informative, and currently available question answering APIs tend to just do simple, factual questions and answers. IBM researcher Bill Murdock believes that IBM Watson could handle Quora questions with a little re-tuning; however, IBM Watson does not currently offer any API. Furthermore, since Quora has no API yet, it would be more challenging to post answers back to Quora.

1 Comment • Jan 23, 2011

= = =

 
http://imified.com is a good place to start for the transport mechanism ..

Check http://pandorabots.com for the intelligence ..

You can find out how to connect Imified with Pandorabots here http://viewer.zoho.com/docs/urlview.do?url=http://alicebot.org/documentation/ThirdParty.doc ..

1 Comment • Jan 21, 2011

= = =

 
That depends on what purpose you want it for, enterprise, small business, or personal. There are quite a few companies offering customer service solutions for substantial companies. There are also a good number of qualified consultants who can implement this kind of solution for small business. And, there are do it yourself hacks available online for the hobbyist. It would also depend on what country you are in, not to mention what language your customers speak, to determine the best solution. ;^)

Comment • Jan 20, 2011

= = =

 
Certainly, people will be able to make a machine with self-awareness; however, that does not mean it will have human self-consciousness. I imagine self-awareness in a machine would be able to examine and modify its own coding, if not its own hardware.

Comment • Jan 20, 2011

= = =

 
Its worth re-examining the IBM Deep Blue (chess computer) story to compare for potential effects .. You can read more about it here:


There is also a good 4 part documentary available on YouTube about Kasparov versus Deep Thought:


See also:


Comment • Jan 19, 2011

= = =

 
Check out the new Denise virtual assistant from guile3d at http://guile3d.com/en ..

1 Comment • Jan 19, 2011

= = =

 
Check out the OpenCog Foundation at http://opencog.org .. 

Ben Goertzel and Hugo de Garis are real characters .. 

See their: 

A world survey of artificial brain projects Part I: Large-scale brain simulations (2010)

and 

World survey of artificial brains, Part II: Biologically inspired cognitive architectures (2010)

Comment • Jan 19, 2011

= = =

 
Two related issues:

1) Better to start with dolphins:

Scientists say dolphins should be treated as 'non-human persons'


2) Issues of agent abuse must be addressed:


= = 

See also Isaac Asimov's Three Laws of Robotics:

Comment • Jan 19, 2011

= = =

 
Real world questions tend toward messy and rambling (like asking three questions in one, and more often than not ungrammatically, if not misspelled or even in SMS jargon), which makes it really difficult for question answering machines. Both WolframAlpha and TrueKnowledge have question answering APIs available today, but seem to only be able to answer brief factual questions. One of the problems with learning machines is that they tend to become degraded when released into the wild, because people will drag them down to the lowest common denominator. So in answer to your question, I would say Watson would be able to answer only a small percentage of questions on Quora.

Comment • Jan 19, 2011

= = =

 
NYT put #ibmwatson hardware at $1 million http://www10.nytimes.com/2010/06/20/magazine/20Computer-t.html ..

#Power750 32core goes for $350 grand  http://www.itjungle.com/tfh/tfh081610-story01.html ..

If 32 core Power 750 retails for $350 grand .. then 2,880 core #IBMWatson would cost you $31.5 million .. http://www.wired.com/epicenter/2011/01/ibm-watson-jeopardy/ ..

2 Comments • Jan 17, 2011

= = =

 
Who can say what is really "normal" in AI? The UIMA natural language architecture could be said to be somewhat standard. The metalearner confidence scoring is an interesting twist. Without a doubt though, its the supercomputer hardware that gives it the edge, doing in seconds what it would take mere mortal machines to do in hours.

Comment • Jan 16, 2011