Analysis and evaluation of query reformulations in different task types

Abstract

Users engaged in information search often reformulate or modify their queries. This paper reports on an investigation of how task type and task situation influence users' query reformulation behavior. A controlled experiment was conducted with 48 participants, each working on six web search tasks classified into three types according to the task structure: Simple, Hierarchical and Parallel. We developed a taxonomy of query reformulation and used an automated method to detect the reformulations. Our results showed that Specialization was most frequently used in Simple tasks, and Word Substitution was most frequently used in Parallel tasks. After visiting and saving a useful web page, Generalization was less likely to be used while New query was more likely to be used. We also found that the effectiveness of each query reformulation type varied in different task types. The results of this study demonstrate the effect of task type on users' query reformulation behavior and have implications for the design of query suggestions that are offered to users during searching.

Age-related Differences in the Content of Search Queries when Reformulating

“…People more familiar with the topic of search were also found to use less queries, more relevant keywords per query, select more relevant results and were more successful in solving tasks, especially the difficult ones [27,30]. Choice of reformulation strategy employed was also shown to be influenced by the type of search task [24], however, this and the earlier studies did not incorporate any age-related differences. The two main issues that this paper addresses are:…”

“…Therefore, if Q i and Q i+1 are two consecutive queries, Q i+1 is a generalization of Q i if Q i and Q i+1 have at least one query term in common, Q i+1 contains fewer or equal number of terms than Q i and/or Q i+1 contains some new terms not present in Q i [24]. For the example task in the difficult category discussed above, if the queries are "niersteen pijn in rug (kidney stones pain in the back)", "branderig gevoel bij plassen (burning sensation when urinating)", "blaasontsteking (urinary infection)" then the number of times generalization strategy is applied is 3.…”

“…We also had other types of reformulations in our data such as spelling corrections (58), repeating same query again (7), converting a query string into its plural form (3), use of special operators such as AND or double quotes (2), reordering words in a query (2) which were dropped from the analysis as these were not many compared to the total number of reformulations we had (together 10.18%). We used a slightly modified version of generalization and specialization as defined in [24].…”

“…Therefore, if Q i and Q i+1 are two consecutive queries, Q i+1 is a specialization of Q i if Q i and Q i+1 have at least one query term in common, Q i+1 contains more terms than Q i and/or Q i+1 contains some new terms not present in Q i [24].Taking the same example task as above, for the queries in the first case, the number of times specialization strategy was applied is 0 as all three queries were generated by using completely new terms. However, for the queries in the second case, the number of times specialization strategy is applied is 1 as the third query is generated by adding additional terms (weight loss) to the second query.…”

Automatic Suggestion for PubMed Query Reformulation

“…The subcategory 'complete change' in the table, which is the most popular (23.9%) but not dominant, means that the revised query is completely different from the original query. To deal with this case, we might need to understand users' behaviors [19,20] and goals [21] in information retrieval, which are out of the scope of our research. Furthermore, around 22% query pairs in the 'others' category are reformulated with the combination of 'permutation' and one of the three basic reformulations.…”