4) Bates: The design of browsing and berrypicking techniques for online search interface.

In "The design of browsing and berrypicking techniques for online search interface," Bates' discuses a new model of interacting with IR systems, once which Bates' argues is closer to humans' natural method of searching. Unlike the classic model, berrypicking takes into consideration "the nature of the query, the nature of the search process, the range of search techniques used, [and]the information 'domain' where the search is conducted" (Bates 1989, pg. 409). This shift from a system-central model to a user-central one was revolutionary.

A user's information need may not be articulate at the start of the query, or never fully articulated at any point. According to Belkin et al. in "Anomalous states of knowledge as a basis for information retrieval," a search beings when a person realizes s/he has an ASK, or gap in his/her knowledge that can be resolved through communication (Belkin 1982, pg. 62). It is important to note that Belkin does not include anything about the query being formed or articulated. This connects also to Taylor, who argues in "Question negotiation and information seeking in libraries" that it is not until the last stage of the search process that a query is developed. Despite this imprecise information need, the classic model of information retrieval was static and rigid: once an information need was translated into a query, the system retrieved a document representation that best matched the query. According to Bates, this single result system is too constrictive. Rather than confirm one's search need to the limitations of the system, Bates discusses a model that allows for the query to evolve as the search proceeds; this she calls "berrypicking," in reference to people's method of literally picking berries off a bush. This connects to Bates in another way: a "berry bush" is like a bunch of ASks; every new cluster of berries that the user finds might answer a part of the ASK or lead to a new ASK.

She differentiates this from "browsing": browsing is part of berrypicking, not the entire method. Berrypicking in manual environments includes browsing, as well as footnote chasing (in which a person looks up the sources that the author used), citation searching (to find other articles that use the same source), searching abstracts and indexes, and author searching (to find other articles written by the author)(Bates 1989, pg. 412). Bates argues that these strategies could also be used when searching with in IR system; time has proved her correct and with the web at searchers' fingertips, we can now utilize berrypicking during our online searches. Hyperlinks allow users to jump from document to document, adapting one's query as one searches through citations on Google Scholar, then through wiki articles, then through references. This way even if a person's information need isn't directly resolved, it may be narrowed down into a searchable query.

Though this method may be inconsistient compared to the more rigid search structures of classic IR retrieval systems, it harkens back to the quote that Taylor opens "Question negotiation and information seeking in libraries" with: "You should be sloppy enough so that the unexpected happens, yet not so sloppy that you cannot figure out what happens after it has happened" (Eiduson, in Taylor 1968, pg. 28).