Detection of abnormal user click activity in a search results page
The present invention provides for the detection of abnormal user behavior for a query session of an electronic search engine. A query session is initiated upon receipt of a user search request that includes one or more search terms. The search engine, in accordance with known search technology, generates a search results page that includes various hyperlinks, including for example web content hyperlinks, page navigation hyperlinks and advertising hyperlinks. Tracking user activities generates the clickstream associated with the search results page. The present invention determines a probability score for the clickstream and then this score is normalized. A comparison of the normalized probability score with other normalized probability scores for similar query sessions determines of the normalcy of the query session.
1. A computer-implemented method for detecting abnormal user behavior for a query session of an electronic search engine through determining clickstream data by tracking user click activities associated with a search results page generated in response to a user search request including a search term, the method comprising:
electronically determining a conformance score for the clickstream data based on existing clickstream data for one or more similar query sessions, the conformance score representing normal user behavior;
normalizing the conformance score that includes a plurality of tracked user transitions to generate a normalized conformance value based on an event count for the query session;
mapping the clickstream characteristics onto a probability score using univariate or multivariate models; and
comparing the probability score with the probability scores for the one or more similar query sessions to determine if the query session is abnormal.
2. The method of claim 1 further comprising:
if the query session has a low probability score relative to the one or more similar query sessions, identifying the query session as abnormal.
3. The method of claim 1 further comprising:
if the query session has a normal or high normalized probability score relative to the one or more similar query sessions, identifying the query session as normal.
4. The method of claim 1 comprising:
determining the conformance score for the clickstream data by comparing the clickstream data against a Markov chain, wherein the Markov chain is generated on the basis of one or more similar query sessions.
5. The method of claim 4 wherein the Markov chain relates to the user click activities for at least one of the web content hyperlinks, page navigation hyperlinks and advertising hyperlinks.
6. The method of claim 1 wherein the step of comparing the normalized probability score includes using at least one of a univariate and multivariate outlier detection model.
7. The method of claim 1 , wherein the step of calculating the normalized probability score includes:
determining a normalized vector of events in the clickstream plus the conformance score;
determining a mean vector across the one or more similar query sessions;
determining a covariance matrix for variables in vector; and
calculating the probability based on the difference of the product of the covariance matrix and the mean vector from the normalized vector.
8. The method of claim 1 comprising:
if the query session is determined abnormal, conducting further analysis of the clickstream.