Why the Big Search Engines are Not Exporatory Search Engines

Monday, May 16, 2011

A few weeks ago we discussed a monograph by M. Wilson, et al., which elucidated the failings the modern search engines when it comes to exploratory search. Wilson and associates concisely characterize exploratory search as revolving around needs that are "open-ended, persistent, and multifaceted, and information-seeking processes that are opportunistic, iterative, and multitactical." Back in March, the Helioid team launched the Glog, a tool meant to bring to light the cases in which Google search results fall short of satisfactory. Keeping in mind the arguments made by Wilson in company, perusing the examples given in the Glog, it's not hard to see the specific ways in which Google searches insufficiently address the needs of exploratory searchers.

The last few months have seen dramatic political upheavals rolling across the Arab world and North Africa, culminating in the topping of oppressive autyocratic regimes in Egypt and Tunisia, as well as ongoing rebellions and uprisings in Libya, Syria, Yemen and Bahrain. Displayed on the Glog, we see a series of searches aimed at exploring the nature of these events from the standpoint of an interested but uninformed web searcher. The searches "protests arab nations", "arab protests origins" and "egyptian protests background" encapsulate an attempt at the kind of exploratory search described by Wilson, et al. An uninformed web searcher likely will not know where to start with a more specific query and is more lkely to start by casting a wide net with a broad, open-ended query like "protests arab nations." The results for this query are largely a smattering of the most popular news and reference articles on the protests, but for an underinformed searcher attempting to suss out a coherent overall picture of what all the buzz has been about the set of articles returned is of little use. Even switching to Google's timeline feature presents a laborious task for such a user, particularly when they are unable to easily narrow the focus of the timeline to threads around particular types of articles they are more likely to be interested in (eg. overviews of protestors' demands and motivations, or the development of the international community's response).

The results returned for the queries "origins arab protests" and "egyptian protests background" similarly illustrate a difficulty of attempting to start from an open ended, fairly broad query and drill down to any desired level of detail. The results are dominated by popular articles directly related to the queries, but particularly for a user who is starting out unsure of what specifically they are looking for, these results provide only a vague picture of the complex political atmosphere that led to the protests. In order to support exploratory search, a search engine would need to provide users submitting such open ended queries with more diverse cross sections of the topics and types of pages related to their query, as well as a method of easily and iteratively drilling down around particular topics of interest.

Current Search Interfaces are Inadequate for Exploratory Search

Tuesday, April 05, 2011

The monograph From Keyword Search to Exploration: Designing Future Search Interfaces for the Web [1] by M. Wilson et al. lucidly reviews the current state of search interfaces that move beyond the traditional list of results. We've included some choice quotations below.

On why we need better search interfaces and systems:

Even more recently, though, researchers have identified just how inadequate the familiar keyword search paradigms, provided by environments such as Google and Bing (Microsoft's search engine), might be for users who need to do more than just find a website that answers a factual question. (p.9)
Exploratory search scenarios are characterized by needs that are "open-ended, per- sistent, and multifaceted, and information-seeking processes that are opportunistic, iterative, and multitactical" [2]. [...] It is plain to see that a search interface needs to provide more than a simple keyword search form to support users in applying such strategies. (p.10)
 

On problems with new search interfaces:

[After evaluating textual and non-textual layouts] the conclusions were that web search results lack "1)... a natural spatial layout of the data; and 2)... good small representations," which makes designing effective visual representations of search results challenging. 3
 

Additionally, here are links to some of the mentioned projects we found interesting and unique.

  • SIMILE Widgets: Free, Open-Source Data Visualization Web Widgets
  • Sii: Search interface inspector
 

[1] Max L. Wilson, Bill Kules, m.c. schraefel and Ben Shneiderman (2010) "From Keyword Search to Exploration: Designing Future Search Interfaces for the Web", Foundations and Trends(R) in Web Science: Vol. 2: No 1, pp 1-97. http://dx.doi.org/10.1561/1800000003

[2] R. W. White and R. Roth, Exploratory Search: Beyond the Query-Response Paradigm. Morgan & Claypool, 2009.

[3] W. Rivadeneira and B. B. Bederson, A Study of Search Result Clustering Interfaces: Comparing Textual and Zoomable User Interfaces. University of Maryland: HCIL, 2003.