On using user query sequence to detect off-topic search

Abstract

Retrieving off-topic documents to a user's pre-defined area of interest via a search engine is potentially a violation of access rights and is a concern to every private, commercial, and governmental organization. We improve content-based off-topic search detection approaches by using a sequence of user queries versus the individual queries. In this approach, we reevaluate how off-topic a query is, based on the sequence of queries that preceded it. Our empirical results show that using the information from the queries in a given query window, the false alarm rate is reduced by a statistically significant amount.

Improving classification based off-topic search detection via category relationships

“…Recall measures the rate of detection, while precision is defined as the ratio of the cases detected correctly as off-topic to the total of the true and false detections. Similar to [11], we use true negatives to calculate precision and recall, as the objective of our method is to decrease false positives and increase true negatives. The F1-measure is a harmonic mean of precision and recall.…”

“…User profiles are of interest to many areas of research, including personalization [15,16,18], peer-to-peer information retrieval [9], malicious data access detection [4], and off-topic search detection [2]. The user profile may be represented in various ways, including (but not limited to) bag of words derived from persumably valid user queries [5,6], bag of words that map to the legitimate user behavior or organizational tasks [7,11], or topics (categories) mapping to the legitimate user interests [13,14]. Similarly, in this work we assume that the user profiles consist of one or more topics based on user's interest.…”

“…The user issued queries (as in [5,6]) or documents retrieved by the query (as in [7,11] and our current approach) are compared to the user's profile. The similarity between the query and the user's profile or the retrieved documents and the user's profile is measured.…”

“…Retrieved documents are clustered in [7] for detecting off-topic search and it is further refined in [11], where segments of queries are grouped Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.…”

“…Content-based off-topic search detection systems are used to gather evidence that fraudulent information access has occurred. Furthermore, the record of off-topic searches can be used to make informed decisions about the subsequent user's queries, such as in [11].…”