A DeepMind Could Improve Google Search
December 16, 2014
DeepMind was invented by London-based genius Demis Hassabis to teach computers how to master complex tasks. He later taught the machines to play classic videogames, which caught Google’s attention and they bought DeepMind for $650 million. Technology Review looks at how the new technology can improve Google in, “Demis Hassabis, Founder of DeepMind Technologies And Artificial-Intelligence Wunderkind At Google, Wants Machines To Think Like Us.”
The article acts as a brief biography of Hassabis, highlighting his intelligence program. Computers programmed with the software were told to play Atari games, but were not programmed with any of the rules. Through trial and error the computers mastered the games through reinforcement learning.
“Artificial intelligence researchers have been tinkering with reinforcement learning for decades. But until DeepMind’s Atari demo, no one had built a system capable of learning anything nearly as complex as how to play a computer game, says Hassabis. One reason it was possible was a trick borrowed from his favorite area of the brain. Part of the Atari-playing software’s learning process involved replaying its past experiences over and over to try and extract the most accurate hints on what it should do in the future.”
Now called Google DeepMind, the team of seventy-five people work in London to apply the technology to all of Google’s products. While learning how to apply AI to Google, Hassabis also dreams of new ways it can be used for bigger and better projects. Until then they’re still playing Atari games.
Mr. Hassabis, start applying DeepMind to search.
Whitney Grace, December 16, 2014
Sponsored by ArnoldIT.com, developer of Augmentext
Debing, Bada Gone: Facebook Search
December 13, 2014
I am not too keen on Facebook. I get odd ball friend requests to our automated Beyond Search account which baffles me. Who wants to be friends with a script.
Nevertheless, I read “Facebook Dumps Microsoft Web Search Results.” The Facebookers have to get control of information access to their content. Note I said “their”, not your content, gentle reader.
The write up, which is a foundation, asserts:
Facebook Chief Executive Mark Zuckerberg has flagged search as one of the company’s key growth initiatives, noting in July that there were more than 1 billion search queries occurring on Facebook every day and hinting that the vast amount of information that users share within Facebook could eventually replace the need to search the Web for answers to certain questions.
Yep, the Googlers at Facebook know there is money in them thar search results.
Debing, bada, gone.
Stephen E Arnold, December 13, 2014
Big Deal for Dassault but Where Is Search?
December 12, 2014
After Dassault Systèms bought Exalead in 2010, we were eager to learn what they would do with one of our favorite search systems. We waited. And waited. Now, as Dassault pursues bigger and better deals around the world, search may have fallen by the wayside. In fact, the company seems to be focused on its 3D technology at the moment. Business Wire reveals, “Dassault Systèms Signs Research Agreement with the Food and Drug Administration for Its ‘Living Heart Project’.” If they must shove search aside, at least they’re working on something that should be good for medicine. The press release states that Dassault:
“… has reached a significant milestone in its project aimed at driving the creation and use of simulated 3D personalized hearts in the treatment and diagnosis of heart diseases and medical device development. Powered by Dassault Systèmes 3DEXPERIENCE platform’s realistic simulation applications, The ‘Living Heart Project’ announced in May of this year, has rapidly moved its first realistic 3D heart simulator into beta test, validated the efficacy of a device and has surpassed 30 contributing member organizations.
“As a key step of this initiative, Dassault Systèmes has signed a five-year collaborative research agreement with the United States’ Food and Drug Administration (FDA) which will initially target the development of testing paradigms for the insertion, placement and performance of pacemaker leads and other cardiovascular devices used to treat heart disease.”
Dassault’s deal with the FDA is a five-year collaborative research agreement. The team is working closely with a long list of cardiologists, medical device companies, and academic researchers. We applaud programs that promise better medical outcomes, especially for a condition as widespread as heart disease.
Still, we can’t help being a bit disappointed that search seems to be an afterthought, if that, for Dassault. Exalead’s CloudView is a search platform we felt was on the right track. Exalead was launched in 2000 and is based in Paris. Dassault Systèms is located in Vélizy-Villacoublay, France. The company has a history of snapping up other businesses, having acquired 11 companies since its founding in 1981.
Cynthia Murrell, December 12, 2014
Sponsored by ArnoldIT.com, developer of Augmentext
MEGA HOPEX Platform Integrated with dtSearch
December 11, 2014
The article titled dtSearch Engine Help MEGA Customers Gain Instant Access to Vital Information on BWW explores the integration of the dtsearch engine into MEGA’s HOPEX platform. The platform is touted for its ability to unify complex enterprise and offer comprehensive information. The demands on MEGA’s repository only grew as businesses need additional documents in order to meet GRC requirements. MEGA called on dtSearch to aid in their ability to manage search across the ever-increasing volume of information. The article explains how the two programs will work together,
“As the volume of these ‘attached’ documents grew, MEGA asked dtSearch to provide its powerful search engine to help customers instantly explore this large quantity of unstructured textual information. The dtSearch Engine will provide users with a ranked list of hits based on their search terms, helping them find information faster. This is an especially important capability as businesses engage in transformation programs to improve performance and profitability, and connect with customers in new ways.”
The president of dtSearch Corp, David Thede, is quoted in the article praising the abilities of HOPEX in synthesizing the complex difficulties presented by the modern standards for enterprise architecture. dtSearch Engine is also recognized for its support of a range of databases and its ability to search instantly across terabytes of data.
Chelsea Kerwin, December 11, 2014
Sponsored by ArnoldIT.com, developer of Augmentext
Baymard Institute Announces E-Commerce Search Issues, EasyAsk Standing Nearby with Solution
December 10, 2014
The article on Digital Journal titled Baymard Institute Study Finds Major Problems with Search on Leading E-Commerce Sites considers the study’s findings that “essential e-commerce search capabilities” are missing from many sites, creating an obstacle in their potential for online sales. Poor test results included low tolerance for misspellings, a lack of support for search based on certain produce features, and a lack of support for thematic searches. Reportedly the study broke down search into twelve categories and EasyAsk, the site search “solution” provider jumped into the ring with a white paper on how to improve search and raise sales. The article explains,
“[The] white paper, Improving E-Commerce Search to Meet the Needs of the Modern Shopper, identifies specific solutions for the problems identified in the Baymard Institute Study. [It] also shows examples of how EasyAsk customers such as The North Face, Oya Costumes, InkJet Superstore, and Travers Tools have delivered an engaging search experience as identified in the Baymard Institute report. “The inadequacies of traditional, outdated keyword search engines are prominently displayed in the Baymard Institute Benchmark Study,” said Craig Bassin, CEO of EasyAsk.”
The immediacy of their response and their collaboration in letting organizations freely download part of the study might raise some eyebrows. The article does not go into detail on who exactly sponsored the study, either.
Chelsea Kerwin, December 10, 2014
Sponsored by ArnoldIT.com, developer of Augmentext
Two Companies Offering Unified Search
December 9, 2014
I am confused. What exactly differentiates some of the vendors offering “unified” search? Another question is, “Are these many functions explicitly designed to deliver outputs that reflect real time content analytics on collected text, images, and videos?
I noted this Sinequa diagram on Twitter (http://bit.ly/12IXmnh):
A day ago I was preparing a short report for a client and came across this diagram for the Attivio active intelligence engine which delivers unified access:
The similarities are quite interesting. What came first, keyword search or the repositioning of search as an application that performs is like a giant Microsoft Office solution?
How do these two companies compare to a next generation information access system (NGIAs)? I see three differences:
- An NGIA system makes search a utility, not a core function or a principal plank in the platform
- The outputs of an NGIA system are designed to make or to trigger a tactical or strategic decision; for example, the output of an NGIA system goes into a system controlling a manufacturing robot
- The purpose of the NGIA system is to deliver a solution that pivots on predictive analytics.
I conclude, therefore, that neither Sinequa nor Attivio are NGIA systems at this time. The companies could engineer their search oriented approach toward an NGIA approach. Attivio has new management to help facilitate this important shift. Sinequa, according to a mid tier consulting firm, is one of the Big Dogs in information processing.
It will be interesting to see how search-centric vendors adapt to the next generation information access market. In my forthcoming monograph on this topic, I explore the substantive differences between search-centric “we do it all” systems and the forward looking NGIA system vendors.
Stephen E Arnold, December 9, 2014
Another Good Enough Challenge to Proprietary Enterprise Search
December 8, 2014
The protestations of the enterprise search vendors in hock for tens of millions to venture funders will get louder. The argument is that proprietary search solutions are just better.
Navigate to “Postgres Full-Text Search Is Good Enough!” This has been the mantra of some of the European Community academics for a number of years. I gave a talk at CeBIT a couple of years ago and noted that the proprietary vendors were struggling to deliver a coherent and compelling argument. Examples of too-much-chest-beating came from speakers representing and Exalead and a handful of consultants. See, for example, http://bit.ly/1zicaGw.
The point of the “Postgres Good Enough” article strikes me as:
Search has became an important feature and we’ve seen a big increase in the popularity of tools like Elasticsearch and SOLR which are both based on Lucent. They are great tools but before going down the road of Weapons of Mass Search, maybe what you need is something a bit lighter which is simply good enough! What do you I mean by ‘good enough’? I mean a search engine with the following features: stemming, ranking/boost, multiple languages, fuzzy search, accent support. Luckily PostgreSQL supports all these features.
So not only are the proprietary systems dismissed, so are the open source solutions that are at the core of a number of commercialization ventures.
I don’t want to argue with the premise. What is important is that companies trying to market enterprise search solutions now have to convince a buyer why good enough is not good enough.
For decades, enterprise search vendors have been engaged in a Cold War style escalation. With each feature addition from one vendor (Autonomy), other vendors pile on more features (Endeca).
The result is that enterprise search tries to push value on customers, not delivering solutions that are valued by customers.
The “good enough” argument is one more example of a push back against the wild and crazy jumbles of code that most enterprise search vendors offer.
The good news is that good enough search is available, and it should be used. In fact, next generation information access solution vendors are including “good enough” search in robust enterprise applications.
What is interesting is that the venture funding firms seem content to move executives in and out of companies not hitting their numbers. Examples include Attivio and LucidWorks (really?). Other vendors are either really quiet or out of business like Dieselpoint and Hakia. I pointed out that the wild and crazy revenue targets for HP Autonomy and IBM Watson are examples of what happens when marketing takes precedent over what a system can do and how many customers are available to generate billions for these big outfits.
Attention needs to shift to “good enough” and to NGIA (next generation information access) vendors able to make sales, generate sustainable revenue, and solve problems that matter.
Displaying a results list is not high on the list of priorities for many organizations. And when search becomes job one, that is a signal the company may not have diagnosed its technological needs accurately. I know there are many mid tier consultants and unemployed webmasters who wish my statements were not accurate. Alas, reality can be a harsh task master or mistress.
Stephen E Arnold, December 8, 2014
Blast toward the Moon With Rocket Software
December 8, 2014
YouTube informational videos are great. They are short, snappy, and often help people retain more information about a product than reading the “about” page on a Web site. Rocket Software has its own channel and the video “Rocket Enterprise Search And Text Analytics” packs a lot of details into 2.49 minutes. The video is described as:
“We provide an integrated search platform for gathering, indexing, and searching both structured and unstructured data?making the information that you depend on more accessible, useful, and intelligent.”
How does Rocket Software defend that statement? The video opens with a prediction that by 2020 data usage will have increased to forty trillion gigabytes. It explains that data is the new enterprise currency and that it needs to be kept organized, then it drops into a plug for the company’s software. The compare themselves to other companies by saying Rocket Software makes the enterprise search and text analytics as simple as a download and then it will be up and running. Other enterprise searches require custom coding, but Rocket Software explains it offers these options out of the box. Plus it is a cheaper product without having to sacrifice quality.
Software usage these days is about functionality and ease of use for powerful software. Rocket Software states it offers this. Try putting it to the test.
Whitney Grace, December 08, 2014
Sponsored by ArnoldIT.com, developer of Augmentext
Attivio Is Now an Oracle Competitor, Not a Search Vendor
December 7, 2014
I read “Oracle Competitor Attivio Promotes Stephen Baker to CEO.” Quite a surprise because Attivio is a search-and-retrieval company with a layer of analytics wrappers. Founded by former Fast Search & Transfer executives, the company ingested more than $30 million in venture funding and now has to generate a return for the stakeholders.
I am not sure if Oracle perceives Attivio as a competitor. MarkLogic, an XML data management vendor, also positioned itself as an Oracle competitor. After hitting a wall at about $60 million and grinding through some new presidents, MarkLogic is keeping a low profile in the markets I track.
Now Attivio may be following this MarkLogic path. Two of the founders of Attivio are moving up. Below Ali Riaz and Sid Probstein is Stephen Baker. Mr. Baker also was a Fast Search & Transfer professional. He worked at RAMP Holdings afar a stint at Reed Elsevier where he was responsible for—wait for it—search.
Attivio co-founder Will Johnson is now the chief technology officer. Mr Johnson is another Fast Search alum. He has worked at GetConnected as—wait for it—a search architect.
My thought is that saying Attivio is a competitor to Oracle is one way to connect semantically with “Oracle.”
But as MarkLogic’s trajectory has demonstrated, there is more to saying a company is “like” Oracle than generating revenue on the scale of Oracle.
Both Attivio and MarkLogic are information access companies. Both want to generate more revenue for their stakeholders. Perhaps a management shift will do the trick.
My view is that if Oracle thought either Attivio or MarkLogic offered a unique, high value service, Oracle would have acquired these companies. Oracle may buy Attivio and MarkLogic. I think the catalyst would be generating and demonstrating rapid revenue growth, expanding margins, and a track record of sustainable revenues. i look forward to a glowing analysis of each firm by IDC’s “expert” Dave Schubmehl in the next month or so. Maybe saying something does make reality change?
Stakeholders want a payback. Management change is a precursor to even more significant activity to benefit those who pumped tens of millions into what may be an old-school approach to information access.
Stephen E Arnold, December 7, 2014
A New Spin on Search: Enterprise Listening Platforms
December 6, 2014
In my Yahoo Alert this morning, I saw an item which puzzled me.
When I clicked on the link, I was shown this item from Yahoo Finance: “Independent Research Firm Ranks Visible Technologies as a Leader in New Enterprise Listening Platforms Report/”
The article from MarketWired informed me:
In addition to securing a Leadership ranking among a pool of 11 enterprise listening software and service providers, Visible received among the second highest rankings in the Strategy category. Its road map was cited as including “self-service research tools and additional automation of client-specific data.” The report also stated that “Visible marries an intuitive dashboard that enables users to uncover insights and refine search with high-quality consulting.”
Yep, search is part of the Forrester “enterprise listening platform” functionality. I must admit that the azure chip consultants will resonate with this phrase. I am not sure what it means. I think I get the search part, but the mashing up dashboards, social media, advanced data processing capabilities, and partnership plans amuses me.
Whatever floats one’s boat and boosts one’s revenues is okay with me. I am not sure what Simpson’s mathematics means but it generates revenues and apparently helps sell newspapers.
Stephen E Arnold, December 6, 2014
-
- Subscribe to Beyond Search
Feature archive
News archive
-
