Cohen presents a practical method of query resolution that, whencombined with a keyword scanner, provides quick processing of massiveBoolean queries. The work discussed applies to selective disseminationof information (SDI) systems. SDI systems and information retrieval (IR)systems represent two complementary models of providing appropriatesubsets of documents. IR systems are characterized by quasi-staticdocument collections to which specific questions are addressed, whileSDI systems run sundry documents against a standing body of queries.Selecting and labeling the documents requires two processes: featurescanning and query resolution, by which the items found are tested todetermine whether they satisfy certain combinationsspecified in the query. The paper presents an effective query resolutionalgorithm, in which large collections of queries are represented bygraphs. The author motivates his choice of the Boolean systems by thefact that they remain easier and more comfortable to handle, and thatmost commercial systems, such as Web systems, continue to use Booleanqueries.
After an introduction, the paper presents models of the queries andthe structure associated with a query. Next, the basic algorithm isdeveloped to accommodate arbitrarily complicated nonrecursive queries.The concepts are strengthened by suggestive examples, and severalaspects of the approach are illustrated using pseudocode. Followingthat, the author discusses implementation issues, especially queuemanagement as a central element in insuring the algorithm speed, andpresents extensions of the basic algorithm. Finally, results from acomplete system embodying the algorithm are reported.
The paper presents valuable research on the routing and selectionof unprocessed data with static queries.