Lee LeBlanc on the future of federated search

Lee LeBlanc is the second runner up in the federated search writing contest. The aim of the contest was to predict the future of federated search. Below is Lee’s bio and his essay, in its entirety.

Lee LeBlanc is Continuing Education and Emerging Technologies Coordinator at SWFLN. Lee’s main interests are “continuous” education, leadership within libraries, and the strategic use of emerging technology in libraries.

Lee is one semester away from a Masters of Information and Library Science with concentrations in Information Architecture and Technology. Lee’s work career has somehow always revolved around information science. Having a librarian as a mentor after he got out of the military and worked through his undergrad, he developed a deep appreciation for what libraries could do for individuals and their communities.

Lee is an invited blogger for Tametheweb.com and also posts on Bibliodox. Lee likes microblogging for his personal life. Flickr is where Lee hangs out most online -if you search for Lee on other popular sites, you may find him there too.

The Future of Federated Search: Muriel doesn’t search, but DFAST does, by Lee LeBlanc

As her red Tesla Roadster winds around a curve, Muriel ticks off the items on her to-do list. Suddenly, she gets a text message:

Before Muriel enters her virtual search terrascape, how did we arrive at Dynamic Federated Autonomous Search Technologies (DFAST)? Our information seeking behaviors will come to be shaped by the information we seek. Devices and the access channels we seek information through will further define our search behaviors. The computer is only one of these devices; interaction search technologies another.

In 1995, a user expended time searching; in 2035, a user spends precious time thinking -differently. The days of sitting in front of a dumb search box are over. Users no longer pound the keys in frustration getting zero results or billions or results. How will this happen?

A user executes a search. Then, gets the results back incrementally. Or they wait until a certain block of time has passed and get results back. Depending on search complexity, DFAST uses an algorithm to quickly analyze the search term. Next, the user gets feedback on how long the search takes to generate results. DFAST additionally prompts the user for refinement of their search should it prove too complex for the system. If this is the case either a search avatar or a human search agent can be contacted to talk about the search process. DFAST figures this out; the user does not.

What’s different about this? Search tools changed dramatically in early 2000s. The user no longer spends time on dumb, blind, arcane search tools. Time is spent exploring their ideas, talking with colleagues, writing, re-writing and modifying their ideas. They’ll even have leisure time returned to them in a world of ever shrinking hours. Time on research was wasted by learning how to interact with dumb search tools. DFAST is an augmented search entity now complimenting the human mind. DFAST does so by working when we do not want to.

Users are never concerned with what databases or systems DFAST negotiates on their behalf. Unless they want granulize their search process, which is a sliding scale from “Go fetch this” to any search command protocol, combined in anyway. From Dialog to Natural Language a user can construct a search term like this:

“I need results from Years=1995,2015,2035
words “information drift” “Federated Search” “Sol Lederman”
w/5=goog*,search*,Federat*
not=Microsoft,Apple,
+(sort academic articles top 100+pull top 100 cited by*allauthors*)+(sort video top 100)+(sort citations top 100)+(sort full text)
+(requestILL=all results with keywords in abstract)
+(sendto:leeleblanc,results to drop.io,rss=leeleblanc,mms=leeleblanc,emailleeleblanc,http=leeleblanc) -(minus=trade publication -advertisements -paid results -blacklists -minus 0-authority ranking)

Future generations will look back on the birth of Federated Search as a thousand monkeys trying to type out Macbeth. While noble monkeys might pull that off with infinite time, Federated Search would not have made such a remarkable evolutionary leap without these new technologies. Even in infinite time, typical search could never match DFAST.

In the 1990s, the Deep Web lay hidden as an ethereal information source. In the early 1990’s, we did not have tools yet to see into this medium; nor understand the social thought-mesh driving information seeking. Proprietary databases had yet to be unlocked. The true information explosion built like a sleepy volcano. What we had in 1990s-2000s: merely a premature information pop. The big bang was intelligent Federated Search built upon many software and hardware technologies. From the 1990s into 2030 we morphed into the realm of Dynamic Federated Autonomous Search Technologies. DFAST mates artificial intelligence with human intelligence. DFAST aggregates, allocates, and manages many of the laboriously intense information seeking tasks. DFAST is an ultramarathon runner, power lifter, martial arts master and erudite scholar.

Future generations saw DFAST emerge search bots, avatars, and proprioceptively intelligent networks.  The choice for interaction mechanism was theirs. Search automatons tirelessly search for us. The search box will retain some of its current form. The box, if the users choose to use a search box, is activated by our queries. Here’s where the underlying structure will change.

A car may still have four wheels in the future; its engine though, has yet to be designed. A Tesla Roadster is considered a novel antique. DFAST agents act on our behalf going to other federated search systems, into the deep reaches of ethereal search mediums.  They negotiate the databases without our knowing what databases they’re going to; they handle authentications. Verifying who you are is the work of the system –not the user. Your avatar or bot is your proxy removing you from login purgatory.

DFAST does the work of the actual search. Our time lies in traveling through search results.  Hours are poured into thinking about what to do with results.  In the past, we spent a lot of time looking at our searches thinking about our searches, and conducting our searches. 

While some of that still takes place, the majority of our search time will not be spent searching. We get a little creative time back to spend on other activities. Search became a process that continuously runs in the background for us, generating new results, new data. The search results are visualized as Edward Tuft information packages (TIPs) or citation-links with profiles drawn from the social web. These are just a few of the ways search data can be viewed.

The 2030’s versions of the iPhone and G1 project into space. Using a helioiPhone, the user enters the visual search world. Using light and air, users walk into their search worlds to see and haptically interact with the results. No longer will you be confined to a keyboard and screen. While the search may start in a simple box, virtually projected worlds from handheld devices are entered to explore the results. We walk about these “search terrascapes” our search agents created for us. When they bring the results back, much like Marco Polo from distant lands, they present us with our results. The network, the tools, and the user work together exploring deeper into the areas of search. Doing so helps a researcher find what may contain the ideas of future research. Virtual worlds are created to store and easily share from anywhere. All of these searches populate a virtual information universe.

The DFAST search agents over time look for patterns in our search behavior. Our agents access a database of other users’ searches. Combining this anonymized data as two results sets will allow the agents to show us areas of improvement for our own search skills and search strategies.  Search is more than semantic.  Search lives its own toiling life. The searches will be ongoing, tireless, and relentlessly refining the algorithms producing more accurate results. The agents further stand sentry for new information channels. Then, devise methods to pipe new sources into current searches.  We see these agents continuously working in the background while we think more about how to use these search materials.

The massive increase in this richly layered DFAST infrastructure came from several ideas:

• Searching is an ongoing, deep, semantic activity augmented by future, unknown search results coupled with mining personal search histories -known and anonymous
• Searching is down through disparate datasets mapped for location sensitive results
• Searching becomes smart: not only do users execute queries in search engines, items like what device is being used is noted and custom results confirm to these expectations
• Aggregated results are ranked by more criteria exposing the connections or showing the gaps among multiple disciplines
• Searching shows tag clouds and keywords drawing from the social web
• Search notifications are set on last search. Others’ search strategies and search words are recompiled
• Shibboleth or other credential systems negotiate access

As we finish, we return to Muriel immersed in an interactive world watching a fascinating archived video. She is puzzled by how people used to stare at static screens no bigger than 20 inches. “It’s no wonder”, she says turning to her avatar, “that no one could find anything. Looking through the eye of needle is no way to search.”

Tags: federated search