Information Retrieval System

An information retrieval (IR) system is (increasingly) a computer-based research tool. The purpose of IR systems is to provide references to pieces of information related to the particular inquiry. Traditional equivalents to IR systems include card catalogs and printed volumes of abstracts or annotated bibliographies. According to F. W. Lancaster, as quoted in the introduction to C. J. van Rijsbergen's Information Retrieval, “An information retrieval system does not inform (i.e., change the knowledge of) the user on the subject of his inquiry. It merely informs on the existence (or nonexistence) and whereabouts of documents relating to his request” (van Rijsbergen, 1979, n.p.).

Information retrieval is a term with a long history and a changing meaning. In the past, an IR system was understood to be a closed information system that offered in-depth information on a very specific topic or area. The system would usually include a mixture of complete documents, abstracts, and bibliographic references. The closest current example of that sort of system that exists today would be a non–Internet-connected information kiosk.

With the arrival of the Internet, closed systems have become fewer, and with documents increasingly being created using computer-based systems, more documents have become available online.

Distinctions are further compounded by a conversational understanding of what is defined as information. Information retrieval scholars like van Rijsbergen have made a distinction between data retrieval (DR) and information retrieval (IR).

In the introduction to Information Retrieval, van Rijsbergen addressed the attributes that make the distinction relevant. Attributes particular to DR include exact matches, an artificial language used in queries, and sensitivity in error response. The attributes for IR, as expected, are the opposite. IR involves partial or best matches, natural language use in queries, and a lack of sensitivity in error responses.

A Web browser is an interesting example of a piece of software that is capable of handling both DR and IR queries. An example of data retrieval would involve typing in a specific universal resource locator (URL), with the result that the specific page of information associated with that URL appears on screen. An artificial language is being used (the URL), which if not entered correctly, results in either the wrong information or no information at all (matching and error sensitivity).

Still using the Web browser, information retrieval is represented by the use of a search engine (for example, Google). Queries can be structured in natural language (What is the temperature of the sun?), with (at the time of this writing) 2,550,000 results. The order of the results, barring financial considerations, is based on Google's proprietary software (partial and best matches).

Information retrieval presents an interesting problem for Web page creators: How can the error sensitivity be maximized? When the query for the sun's temperature was run, the contents of the 2.5 million Web pages found were not scanned. Rather, each Web page has keyword entries. The keywords are not visible unless the Web page's source code is examined. In the case of the Web page that came up first in the query, the keyword entries

...