Category: Computing & data (Page 83 of 90)

This recap might have the ring of an old news story but these clips are worth repeating until more enterprises get serious about making search work for them, instead of allowing search to become an expensive venture in frustration. Enterprise Search Europe, May 14-16, 2013, was a small meeting with a large punch. My only regret is that the audience did not include enough business and content managers. I can only imagine that the predominant audience members, IT folks, are frustrated that the people whose support they need for search to succeed were not in attendance to hear the messages.

Here are just a few of the key points that business managers and those who “own” search budgets need to hear.

On Day 1 I attended a workshop presented by Tony Russell-Rose [Managing Director, UXLabs and co-author of Designing the Search Experience, also at City University London], Search Interface Design. While many experts talk about the two top priorities for search success, recall (all relevant results returned) and precision (all results returned are relevant), they usually fail to acknowledge a hard truth. We all want “the whole truth and nothing but the truth,” but as Tony pointed out, we can’t have both. He went on to offer this general guidance on the subject; recall in highly regulated or risk intensive business is most important but in e-commerce we tend to favor precision. I would add that in enterprises that have to manage risk and sell products, there is a place for two types of search where priorities vary depending on the business purpose. My takeaway: universal, all-in-one search implementations across an enterprise will leave most users disappointed. It’s time to acknowledge the need for different types of implementations, depending on need and audience.

Ed Dale [Digital Platforms Product Manager, Ernst & Young (USA)] gave a highly pragmatic keynote at the meeting opening, The Six Drivers for Search Quality. The overarching theme was that search rests on content. He went on to describe the Ernst & Young drivers: the right content, optimized for search, constant tuning for optimal results, attention to a user interface that is effective for a user-type, attention to user needs, consistency in function and design. Ed closed with this guidance: develop your own business drivers based on issues that are important to users. Based on these and the company’s drivers, focus your efforts, remembering that you are not your users.

The Language of Discovery: A Toolkit for Designing Big Data Interfaces and Interactions was presented by Joseph Lamantia, [UX Lead: Discovery Products and Services, Oracle Endeca]. He shared the idea that discovery is the ability to understand data, and the importance of not treating data, by itself, as having value without achieving discovery. Discovery was defined as something you have seen, found, and made sense of in order to derive insight. It is achieved by grasping or understanding meaning and significance. What I found most interesting was the discussion of modes of searching that have grown out of a number of research efforts. Begin with slide 44, “Mediated Sense making” to learn the precursors that lead into his “modes” description. When considering search for the needy user, this discussion is especially important. We all discover and learn in different ways and the “mode” topic highlights the multitude of options to contemplate. [NOTE: Don’t overlook Joe’s commentary that accompanies the slides at the bottom of the SlideShare.]

Joe was followed by Tyler Tate, [Cofounder, TwigKit] on Information Wayfinding: A New Era of Discovery. He asked the audience to consider this question, “Are you facilitating the end-user throughout all stages of the information seeking process?” The stages are: initiation > selection > exploration > formulation > collection > action. This is a key point for those most involved in user interface design and content managers thinking about facet vocabulary and sorting results.

Steve Arnold [Arnold IT], always brings a “call to reality” aspect to his presentations and Big Data vs. Search was no different. On “Big Data” a couple of key points stick out, “More Data” is not just more data; it is different. As soon as we begin trying to “manage” it we have to apply methods and technologies to reduce it to dimensions that search systems can deal with. Search data processing has changed very little for the last 50 years and processing constraints limit indexing capabilities across these super large sets. There are great opportunities for creating management tools (e.g. analytics) for big data in order to optimize search algorithms, and make the systems more affordable and usable. Among Arnold’s observations was the incessant push to eliminate humans, getting away from techniques and methods [to enhance content] that work and replacing them with technology. He noted that all the camera and surveillance systems in Boston did not work to stop the Marathon bombers but people in the situation did limit casualties through quick medical intervention and providing descriptions of suspicious people who turned out to be the principal suspects. People must still be closely involved for search to succeed, regardless of the technology.

SharePoint lurks in every session at information technology conferences and this meeting was no exception. Although I was not in the room to hear the presentation, I found these slides from Agnes Molnar [International SharePoint Consultant, ECM & Search Expert, MVP] Search Based Applications with SharePoint 2013 to be among the most direct and succinct explanation of when SharePoint makes sense. It nicely explains where SharePoint fits in the enterprise search eco-landscape. Thanks to Agnes for the clarity of her presentation.

A rapid fire panel on “Trends and Opportunities” moderated by Allen Peltz-Sharpe [Research Director for Content Management & Collaboration, 451 Research] included Charlie Hull [Founder of Flax], Dan Lee of Artirix, Kristian Norling of Findwise (see Findwise survey results), Eric Pugh of OpenSource Connections and Rene Kreigler an independent search consultant. Among the key points offered by the panelists were:

• There is a lot to accomplish to make enterprise search work after installing the search engine. When it comes to implementation and tuning there are often significant gaps in products and available tools to make search work well with other technologies.
• Search can be leveraged to find signals of what is needed to improve the search experience.
• Search as an enterprise application is “not sexy” and does not inspire business managers to support it enthusiastically. Its potential value and sustainability is not well understood, so managers do not view it as something that will increase their own importance.
• Open source adoption is growing but does face challenges. VC backed companies in that arena will have a struggle to generate enough revenue to make VCs happy. The committer community is dominated by a single firm and that may weaken the staying power of other search (Lucene, Solr) open source committers.

A presentation late in the program by Kara Pernice, Managing Director of NN/g, Nielsen Norman Group, positioned the design of an intranet as a key element in making search compelling. Her insights reflect two decades of “Eyetracking Web Usability” done with Jakob Nielsen, and how that research applies for an intranet. Intranet Search Usability was the theme and Kara’s observations were keenly relevant to the audience.

Not the least of my three days at the meeting were side discussions with Valentin Richter CEO of Raytion, Iain Fletcher of Search Technologies, Martin Rugfelt of Expertmaker, Benoit Leclerc of Coveo, and Steve Andrews an advisor to Q-Sensei. These contributed many ideas on the state of enterprise search. I left the meeting with the overarching sense that enterprise leadership needs to be sold on the benefits for sustaining a search team as part of the information ecosystem. Bringing an understanding of search as not just being a technological, plug & play product and a “one-off” project is the challenge. Messaging is not getting through effectively. We need strong and clear business voices to make the case; the signals are too diffuse and that makes them weak. My take is that messages from search vendors all have valid points-of-view but when they are combined with too many other topics (e.g. “big data,” “analytics,” “open source,” SharePoint, “cloud computing”) basic concepts of what search is and where it belongs in the enterprise gets lost.