Coniunge Et Impera: Multiple-Graph Mining for Query-Log Analysis
Abstract
Query logs of search engines record a huge amount of data about the actions of the users who search for information on the Web. Hence, they contain a wealth of valuable knowledge about the users’ interests and preferences, as well as the implicit feedback that Web searchers provide when they click on the results obtained for their queries. In this paper we propose a general and completely unsupervised methodology for query-log analysis, which consists of aggregating multiple graph representations of a query log, tailored to capturing different semantic information. The combination is carried out by applying simple but efficient graph-mining techniques. We show that our approach achieves very good performance for two different applications, which are classifying query transitions and recognizing spam queries.
Cite
Text
Bordino et al. "Coniunge Et Impera: Multiple-Graph Mining for Query-Log Analysis." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2010. doi:10.1007/978-3-642-15880-3_17Markdown
[Bordino et al. "Coniunge Et Impera: Multiple-Graph Mining for Query-Log Analysis." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2010.](https://mlanthology.org/ecmlpkdd/2010/bordino2010ecmlpkdd-coniunge/) doi:10.1007/978-3-642-15880-3_17BibTeX
@inproceedings{bordino2010ecmlpkdd-coniunge,
title = {{Coniunge Et Impera: Multiple-Graph Mining for Query-Log Analysis}},
author = {Bordino, Ilaria and Donato, Debora and Baeza-Yates, Ricardo},
booktitle = {European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases},
year = {2010},
pages = {168-183},
doi = {10.1007/978-3-642-15880-3_17},
url = {https://mlanthology.org/ecmlpkdd/2010/bordino2010ecmlpkdd-coniunge/}
}