Leave a comment

Mar 31, 2011 | Reinventing the classic search model

As a result of the Podio Hack Day project I had the opportunity to talk to InboxQ’s CEO Joe Fahrner, and I think there’s a lot of opportunities he and his team are tackling that are worthy of discussion here. From a high level perspective, InboxQ identifies questions made on social networks (in particular, Twitter) and makes it possible to either respond or leverage context to identify the right person to answer that particular question.

Reimagining the classic search model

The idea here is very much counter to existing models for Q&A sites where two major goals are proposed:

By contrast, identifying questions in situ gives the opportunity for finding raw and unadulterated questions – questions that are posed to a certain group of individuals whose relationship to the asker is decidedly unique. Furthermore, it is becoming harder and harder to maintain any decent social presence in any site now because there are so many to participate in – my choice to participate in one (StackOverflow), may mean forgoing the opportunity to engage in another (Quora). Lastly, the responsibility around the question shifts significantly from the system to the user – whereas previously a user would have to identify the right forum the ask the question and to also recognize that a question has not been asked before, the user now has the freedom to ask any question, any time, and anywhere.

Inherently, a result of that is likely to be richer questions and more relevant answers. A good example is a personal one: I am looking for tickets to the TEDxSV conference and posted the same on Twitter:

Reimagining the classic search model

The response I obtained was one that was not only extremely helpful, but indicative of the fact that some knowledge forms are longitudinal.

Reimagining the classic search model

I think what it also promises is the notion that the classic search model (slide from Chris Manning) no longer needs to involve a translation from the mental model to the practical approaches whenever queries are made online. As it is right now, a query is likely to be a compromise from the real intention and it is only through careful algorithms can a system pinpoint the initial user goal.

Reimagining the classic search model

I think what is worth noting here is the fact that question-answering is very much an information retrieval process as it is right now – instead of looking for the document that has the answer (a la Google), though, we are looking for people with the right answer. But what if our perceptions of where to find the right people is limited? Can we not assign algorithms to do that for us? That is perhaps why Vark seemed to have so much promise.

Reimagining the classic search model

Clearly some obvious benefits to the existing model is that it provides an arena where people who have questions and people who want to answer them can meet – Vark can tend to be annoying if we’re not in a mood to answer questions. The current model also allows many individuals to benefit from a single answer, but perhaps that notion may be up for challenge in favor of more contextual, personal, and richer interactions that stem from questions and answers.

I may be exceedingly optimistic, but if anything, I’m hopeful that InboxQ’s approach will make the noisy, daunting, unforgiving web just a tad bit friendlier.

This entry was posted on Thursday, March 31st, 2011 at 1:20 am, EST under the category of Coding. You can leave a response, or trackback from your own site.