[Paper Review] An Effective Clustering Approach to Web Query Log Anonymization

Amin Milani Fard, Ke Wang|arXiv (Cornell University)|Dec 3, 2010

Privacy-Preserving Technologies in Data17 references3 citations

TL;DR

This paper proposes a novel clustering-based approach to anonymize web query logs by modeling queries as unstructured transaction data and applying k-anonymity through clustering and generalization. Experiments on AOL query logs show higher data utility than state-of-the-art methods while preserving privacy.

ABSTRACT

Web query log data contain information useful to research; however, release of such data can re-identify the search engine users issuing the queries. These privacy concerns go far beyond removing explicitly identifying information such as name and address, since non-identifying personal data can be combined with publicly available information to pinpoint to an individual. In this work we model web query logs as unstructured transaction data and present a novel transaction anonymization technique based on clustering and generalization techniques to achieve the k-anonymity privacy. We conduct extensive experiments on the AOL query log data. Our results show that this method results in a higher data utility compared to the state of-the-art transaction anonymization methods.

Motivation & Objective

• To address privacy risks in releasing web query logs, which can re-identify users even without explicit identifiers.
• To overcome limitations of existing anonymization techniques that reduce data utility while ensuring privacy.
• To develop a method that achieves k-anonymity by clustering and generalizing query logs while preserving meaningful patterns.
• To evaluate the proposed method on real-world query log data and compare its utility to existing approaches.

Proposed method

• The authors model web query logs as unstructured transaction data, treating each query as a transaction of keywords.
• They apply a clustering algorithm to group similar queries based on keyword overlap and semantic similarity.
• Within each cluster, queries are generalized using techniques such as replacing specific terms with broader categories or wildcards.
• The method enforces k-anonymity by ensuring each cluster contains at least k queries, preventing re-identification.
• Generalization is applied selectively to minimize information loss while maintaining privacy.
• The approach is evaluated on the AOL query log dataset, using metrics to compare utility and privacy.

Experimental results

Research questions

• RQ1 Can clustering-based generalization techniques effectively anonymize web query logs while preserving data utility?
• RQ2 How does the proposed method compare to existing transaction anonymization techniques in terms of k-anonymity and data utility?
• RQ3 To what extent can query logs be generalized without losing meaningful patterns for research purposes?
• RQ4 What is the impact of different clustering strategies on the balance between privacy and utility?

Key findings

• The proposed clustering-based anonymization method achieves higher data utility compared to state-of-the-art transaction anonymization techniques on the AOL query log dataset.
• The method successfully enforces k-anonymity by forming clusters where each contains at least k queries.
• Generalization is applied efficiently within clusters, minimizing loss of semantic and structural information in the logs.
• The approach maintains meaningful query patterns, enabling continued use of anonymized logs for research.
• Experimental results demonstrate that the method outperforms existing methods in preserving data utility under the same privacy constraints.
• The method is effective even when applied to large-scale, real-world query logs with diverse and complex query patterns.