Query-Task Mapping

Völske, Michael; Fatehifar, Ehsan; Stein, Benno; Hagen, Matthias

doi:10.1145/3331184.3331286

Cited by 8 publications

(10 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previously analyzed methods for mapping queries face a trade-off between accuracy and execution time. The most accurate method uses an inverted index approach based on a BM25 retrieval model; however, its average time per query is much slower than a Trie data structure implementation, which is the fastest method [17]. To address this trade-off, we utilize the Neighborhood Graph and Tree (NGT) approximate nearest neighbor method [9], along with S anд and multilingual query vectors.…”

Section: Mapping Queries To Search Tasksmentioning

confidence: 99%

“…Three datasets have been proposed to evaluate the mapping of new incoming queries: the AOL query log (AOLQL), the TREC query topics (TRECQT), and the WikiHow questions (WIKIHQ) based datasets [17]. For comparison, we use publicly available implementations for the Trie data structure 1 and the BM25 retrieval model 2 with default parameters [17,19]. Experiments run on a virtual machine instance with 8 CPUs of 3GHz and 60GB of RAM.…”

Section: Mapping Queries To Search Tasksmentioning

confidence: 99%

“…Experiments run on a virtual machine instance with 8 CPUs of 3GHz and 60GB of RAM. We compute time per query as the average time for mapping 10 4 queries, while accuracy is measured using a leave-one-out evaluation, independently selecting 100 random queries from the dataset and repeating the evaluation during 50 runs [17]. We test several values of k nearest neighbors for NGT, finding k = 9 as the best performing setup.…”

Section: Mapping Queries To Search Tasksmentioning

confidence: 99%

“…Mining query logs from search engines enable the automatic modeling of search tasks. Precise modeling of search tasks is required for user supporting applications like query term prediction, query recommendations, user modeling based on tasks, personalization in e-commerce, and results ranking [8,11,17], which enhance search engine support for helping users to fulfill their information needs.…”

Section: Introductionmentioning

confidence: 99%

“…The proposed search task identification approach is independent of user identifiers, enabling the modeling of search tasks in user agnostic or personalized applications. We also address the existing trade-off between accuracy and query time [17] that arises when mapping new incoming queries to identified search tasks.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

A Multilingual Approach for Unsupervised Search Task Identification

Lugo

Moreno

Hubert

2020

Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval

View full text Add to dashboard Cite

Users convert their information needs to search queries, which are then run on available search engines. Query logs registered by search engines enable the automatic identification of the search tasks that users perform to fulfill their information needs. Search engine logs contain queries in multiple languages, but most existing methods for search task identification are not multilingual. Some methods rely on search context training of custom embeddings or external indexed collections that support a single language, making it challenging to support the multiple languages of queries run in search engines. Other methods depend on supervised components and user identifiers to model search tasks. The supervised components require labeled collections, which are difficult and costly to get in multiple languages. Also, the need for user identifiers renders these methods unfeasible in user agnostic scenarios. Hence, we propose an unsupervised multilingual approach for search task identification. The proposed approach is user agnostic, enabling its use in both user-independent and personalized scenarios. Furthermore, the multilingual query representation enables us to address the existing trade-off when mapping new queries to the identified search tasks.

show abstract

Section: Mapping Queries To Search Tasksmentioning

confidence: 99%