On Wednesday, February 4th, my colleague Andrew McKay and I had the privilege of speaking at the first Enterprise Search Forum Toyko at the International House of Japan in Roppongi. Attivio was co-sponsor of the event, which was hosted by our partner Basis Technology.

First up: a technical training session for several system integrators. We spent almost two hours detailing the Active Intelligence Engine (AIE) architecture, installing the software, ingesting content (the World Factbook, medal results from the 2008 Olympics and several news feeds), then running a variety of typical and JOIN() queries. It was a terrific session; I particularly enjoyed the Q&A at the end where we talked about analyzing and tuning relevancy, the realities of deploying and operating software in production, etc.

Carl Hoffman, CEO of BasisTech, opened the forum with interesting remarks about the changing of the guard in Enterprise Search. He spoke of the healthy cycle by which legacy players evolve into different companies or are acquired by larger ones, thus leaving room for new, innovative players to enter the market and redefine it. He also spoke of the value of Japan as a vital market, and particularly the sophisticated interest in information access.

Next up was the keynote by Mr. Kazuyuki Ide, Software Research Manager at IDC Japan. Ide-san first talked about the continued, unchecked growth of digital information in the enterprise, noting that although growth in the web was slowing, the rate in the enterprise was actually increasing. He also presented survey results highlighting the continued need for next-generation information access systems to reduce the time spent on searching for and gathering information - 16+ hours per week - versus performing actual analysis (~8 hours per week). Later he detailed new directions in enterprise search including multimedia and image search, relationship analysis and text analytics such as sentiment analysis.

After a short break I presented Attivio's view of the future: unified information access (UIA) to structured and unstructured data. Some of the major points:

• Legacy enterprise search has not broken down silos; rather, it has just created new ones

• BI/DW/reporting systems have reached a steady state

• The challenge now is to do analysis across these two incompatible silos; this is how competitive advantage will be obtained in the coming years

Then I detailed the advantages of UIA in various segments and used customer case studies to illustrate each point. For example:

• In Governance, Risk and Compliance (GRC) applications coverage of content and data is a critical driver. If you analyze only the unstructured text data, you expose yourself to risk on the structured side. I also gave an example of how a stock trading firm was able to move from random sampling of email, at significant cost and effort, to an automated model whereby email messages to external domains that mention companies (or ticker symbols) traded on by the sender in the previous 48 hours are automatically routed for review; the result was a significant cost savings and a measurable reduction in risk.

• For Information Portals, UIA offers the ability to add structured data to the mainstay unstructured content. Examples include survey results, content analytics and complex security/subscription models (which are most easily modeled as structured, related data). Relevancy can also be improved significantly by incorporating transaction data, user rating and tagging, online/offline status, etc. The latter has posed a challenge for many legacy search players who can't update their index rapidly enough to keep up with this real-time flow of information.

• For Decision Support, UIA improves decision making by providing access to more information at the critical moment. For example, in the customer service arena, not having an important piece of information can make all the difference. Survey results have shown that ~68% of customers who churn out of a company's base do so because of indifferent service. I then detailed a real life story of a 401K rollover gone wrong. The rollover was no problem - it took 15 minutes via phone - but the following day a different associate at the firm called back to ask if they could help roll the 401K over. This was both surprising and alarming. When told that the rollover had already been done, the associate indicated that they didn't have access to that information, and apologized. The cost to the firm was significant: the excess call to do the rollover, and then handling a third call to verify that, in fact, the rollover had been done.

Finally I presented the AIE architecture and demonstrated search, entity extraction and faceted browsing on Japanese Wikipedia. We also showed a new version of our baseball demo that combines baseball statistics and news data, JOINed at query time to produce a variety of interesting visualizations, now including the Japanese MLB news feed.

Andrew and I would like to thank all of the attendees and co-sponsors and especially the team at Basis Technology for being such wonderful hosts. We look forward to attending ESF Japan next year!

TrackBack URI for this entry

Comments (0)