Answer Me!

Bing's Sanaz Ahari on System Feedback (2 of 2) - O'Reilly Radar

Bing's Sanaz Ahari on System Feedback (2 of 2) - O'Reilly Radar.

An interview by Brady Forrest  with Sanaz Ahari, Lead PM on Bing. She led the team that developed the categories you see on a Bing web search. The interview was based on the slides from her presentation at [a small search summit for analysts, bloggers, SEO experts, entrepreneurs and advertisers.] The first portion of the interview focused on how the Bing team handles Query level categorization and some of the problems they face. This second portion ... focuses on the systems used to generate the categorization.

I, personally, think that Bing's failure to distinguish subtopics from disambiguations is a mistake.  "Jennifer Lopez" has subtopics related to her perfume, her music, etc, but "Jaguar" is not a unified topic, so saying this one topic has subtopics "stadium", "dealers", etc is just confusing.  Lycos Retriever did this better!

Bing's Sanaz Ahari on Query Level Categorization (1 of 2) - O'Reilly Radar

Bing's Sanaz Ahari on Query Level Categorization (1 of 2) - O'Reilly Radar.

An interview with Sanaz Ahari, Lead PM on Bing. She led the team that developed the categories you see on a Bing web search. The interview was based on the slides from her presentation at the event. ...The first portion of the interview focuses on how the Bing team handles Query level categorization and some of the problems they face. The second portion (up shortly) focuses on the systems used to generate the categorization.

imense® - Autotagger

Interesting approach to photo autotagging.  Doesn't seem to involve geotagging, however.  Pity.

imense® - Autotagger.

Autotagger from imense® provides a consistent approach to keywording, removing the subjective and often error prone nature of traditional human keywording. Image agencies; news agencies; publishing groups and Digital Asset Management providers now have a fast and accurate method of tagging images with dramatic cost savings, resulting in lower operational costs for themselves and their clients.

Technology

Autotagger uses Ontology based visual information processing technology to extract visual features from images. Developed over the last 9 years at our Cambridge laboratories, this unique method allows our algorithms to output thousands of different keywords.

Traditional Content Based Image Recognition (CBIR) techniques are limited in scope to recognize general scenes such as “outside” or “Sports” or recognizing a “Face” within images. The computational difficulty in recognising objects and abstract concepts such as “Happy” or “Joy” has meant alternative systems have been poorly adopted. Through a combination of our Ontology based information processing with our linguistic ontology and using advanced statistical inference techniques, imense® technology breaks traditional CBIR barriers.

The Autotagger platform can now produce the majority of keywords within 1 second, including abstract concepts.

Technology Review: Search Me

Technology Review: Search Me.

Inside the launch of Stephen Wolfram's new "computational knowledge engine."

Breakingviews.com - In Search of Search - NYTimes.com

Breakingviews.com - Encouraging Signs From Some I.P.O.’s - NYTimes.com.

Suddenly, search engines seem to be multiplying. Microsoft is set to introduce Kumo, its latest effort to crack Google’s dominance in search.

Meanwhile, sites like WolframAlpha and Newssift are trying to find safe and profitable niches for specialized searches. Sadly, up to now Google’s supposed rivals have generated more hype than meaningful results. Microsoft’s efforts to take on Google are starting to appear quixotic. It has been introducing search initiatives for years, yet its share has shrunk to less than 10 percent of the market, according to comScore, compared with Google’s 64 percent. Kumo’s new features sound evolutionary, not radical. A more promising approach is to concentrate on niches overlooked by Google.

The new WolframAlpha search engineaims to answer technical and scientific questions. That sounds promising because the site’s founder, Stephen Wolfram, developed the well regarded program Mathematica. WolframAlpha’s method also sounds impressive: the site uses algorithms and interrogates high-quality databases to compute answers to questions. Unfortunately, the results don’t deliver, even in its supposed forte of math and sciences. Type in “second derivative” and the site hasn’t a clue. Biology doesn’t fare better. Type in “dog” and you get an impressive list of statistics on the species Canis lupus. Dogs are indeed a sub-species of grey wolves, but chihuahuas don’t get as heavy as people and have tails almost two feet long.

Newssift, the specialized search site for news introduced by the Financial Times Group, offers a host of helpful-looking methods to drill down on news items. The results, sadly, appear scattershot. These could just be teething pains of new entrants. But failing to hit the ground running makes it easier for Google to catch up and replicate their features. And history doesn’t provide much reason for optimism. Take Cuil. Former Google employees introduced the site last year, but the search engine couldn’t even locate itself at first. It hasn’t caught on. It’s not so much a problem with the new sites’ ideas or their developers’ pedigrees. It’s more that, if it chooses, Google can outgun them so easily. That may become a concern for competition-minded regulators.

ROB COX, ALIZA ROSENBAUM and ROBERT CYRAN

A Brief History of Google Killers | Technologizer

Brilliant, but what would tech journalists do without the "Is <insert company name> the next Google?" or "This new technology <insert technology> understands what you mean!" story templates?

A Brief History of Google Killers | Technologizer.

Fourteen companies that were supposed to do away with the Web's biggest brand. Some of which themselves no longer exist.

By Harry McCracken

Excited About The Wolfram Alpha Launch? Watch It On A Webcast - Seriously! | Singularity Hub

Excited About The Wolfram Alpha Launch? Watch It On A Webcast - Seriously! | Singularity Hub.

In its latest show of innovation, we now learn that Wolfram Alpha will turn its launch into a “community” event. People the world over can watch the Wolfram Alpha team on a live webcast as they try to pull off their ambitious launch. Although the site is officially slated to launch on May 18, starting later today anyone interested can watch the Wolfram Alpha team as they build up for the official launch on a public channel on Justin.tv.

Lifehacker - Task.fm Turns Natural Language Commands into Future Reminders - Reminders

Sounds nice, but natural language querying would be nice, too.  "What do I have next Friday between 10 and 2?", for example.

There's certainly a lot of work on understanding calendar expressions these days.

Lifehacker - Task.fm Turns Natural Language Commands into Future Reminders - Reminders.

Task.fm is a simple web application that turns your natural language commands into email and SMS reminders.

Task.fm takes commands like "meet with Jim tomorrow" or "Replace fish tank filter in 21 days," and converts those commands into future reminders. The reminders will be emailed or sent to you via SMS. The email reminders are free but the SMS reminders require credits with the service—100 messages cost $8. The fee structure isn't outrageous, but we're in agreement that nothing sounds as good as free.

Natural Language Engineering has contents of Special Issue on Interactive Question Answering available for free:

Introduction
	Special issue on interactive question answering: Introduction N. WEBB and B. WEBBER Natural Language Engineering , Volume 15, Special Issue 01, January 2009, pp 1-8 doi:10.1017/S1351324908004877 (About doi), Published Online by Cambridge University Press 22 Oct 2008 Preview Abstract PDF (99 KB) HTML (39 KB) Request Permissions Abstract HTML Article Structure Figures and Tables References
	Papers
	Interactive question answering and constraint relaxation in spoken dialogue systems S. VARGES, F. WENG and H. PON-BARRY Natural Language Engineering , Volume 15, Special Issue 01, January 2009, pp 9-30 doi:10.1017/S1351324908004889 (About doi), Published Online by Cambridge University Press 06 Nov 2008 Preview Abstract PDF (691 KB) HTML (586 KB) Request Permissions Abstract HTML Article Structure Figures and Tables References
	HITIQA: High-quality intelligence through interactive question answering S. SMALL and T. STRZALKOWSKI Natural Language Engineering , Volume 15, Special Issue 01, January 2009, pp 31-54 doi:10.1017/S1351324908004890 (About doi), Published Online by Cambridge University Press 06 Nov 2008 Preview Abstract PDF (1.374 MB) HTML (1016 KB) Request Permissions Abstract HTML Article Structure Figures and Tables References
	Does this list contain what you were searching for? Learning adaptive dialogue strategies for interactive question answering V. RIESER and O. LEMON Natural Language Engineering , Volume 15, Special Issue 01, January 2009, pp 55-72 doi:10.1017/S1351324908004907 (About doi), Published Online by Cambridge University Press 22 Oct 2008 Preview Abstract PDF (374 KB) HTML (290 KB) Request Permissions Abstract HTML Article Structure Figures and Tables References
	Designing an interactive open-domain question answering system S. QUARTERONI and S. MANANDHAR Natural Language Engineering , Volume 15, Special Issue 01, January 2009, pp 73-95 doi:10.1017/S1351324908004919 (About doi), Published Online by Cambridge University Press 28 Oct 2008 Preview Abstract PDF (478 KB) HTML (357 KB) Request Permissions Abstract HTML Article Structure Figures and Tables References
	Follow-up question handling in the IMIX and Ritel systems: A comparative study B. W. VAN SCHOOTEN, R. OP DEN AKKER, S. ROSSET, O. GALIBERT, A. MAX and G. ILLOUZ Natural Language Engineering , Volume 15, Special Issue 01, January 2009, pp 97-118 doi:10.1017/S1351324908004920 (About doi), Published Online by Cambridge University Press 24 Oct 2008 Preview Abstract PDF (198 KB) HTML (251 KB) Request Permissions Abstract HTML Article Structure Figures and Tables References
	Questionnaires for eliciting evaluation data from users of interactive question answering systems D. KELLY, P. B. KANTOR, E. L. MORSE, J. SCHOLTZ and Y. SUN Natural Language Engineering , Volume 15, Special Issue 01, January 2009, pp 119-141 doi:10.1017/S1351324908004932 (About doi), Published Online by Cambridge University Press 06 Nov 2008 Preview Abstract PDF (847 KB) HTML (970 KB) Request Permissions Abstract HTML Article Structure Figures and Tables References
	Adding semantic roles to the Chinese Treebank NIANWEN XUE and MARTHA PALMER Natural Language Engineering , Volume 15, Special Issue 01, January 2009, pp 143-172 doi:10.1017/S1351324908004865 (About doi), Published Online by Cambridge University Press 24 Oct 2008 Preview Abstract PDF (395 KB) HTML (1.187 MB) Request Permissions

BBC NEWS | Technology | Web tool 'as important as Google'

Comments by Nova Spivak (positive) and Boris Katz (negative, surprise!):

BBC NEWS | Technology | Web tool 'as important as Google'.

The tool computes many of the answers "on the fly" by grabbing raw data from public and licensed databases, along with live feeds such as share prices and weather information.

People can use the system to look up simple facts - such as the height of Mount Everest - or crunch several data sets together to produce new results, such as a country's GDP.

Other functions solve complex mathematical equations, plot scientific figures or chart natural events. "Like interacting with an expert, it will understand what you're talking about, do the computation, and then present you with the results," said Dr Wolfram.

As a result, much of the data is scientific, although there is also limited cultural information about pop stars and films.

Dr Wolfram said the "trillions of pieces of data" were chosen and managed by a team of "experts" at Wolfram Research, who also massage the information to make sure it can be read and displayed by the system.