Identification of events of search queries
Techniques for analyzing and modeling the frequency of queries are provided by a query analysis system. A query analysis system analyzes frequencies of a query over time to determine whether the query is time-dependent or time-independent. The query analysis system forecasts the frequency of time-dependent queries based on their periodicities. The query analysis system forecasts the frequency of time-independent queries based on causal relationships with other queries. To forecast the frequency of time-independent queries, the query analysis system analyzes the frequency of a query over time to identify significant increases in the frequency, which are referred to as “query events” or “events.” The query analysis system forecasts frequencies of time-independent queries based on queries with events that tend to causally precede events of the query to be forecasted.
1. A method in a computing device for identifying events of a query, the method comprising:
storing frequency of the query at intervals, the stored frequency for an interval representing number of queries submitted during that interval;
identifying a sequence of intervals in which the frequency satisfies a baseline criterion by comparing the frequency of each interval to a baseline threshold frequency of the baseline criterion;
determining whether a frequency within an identified sequence of intervals satisfies a climax criterion by comparing the frequency of each interval within the identified sequence of intervals to a climax threshold frequency of the climax criterion;
when it is determined that a frequency satisfies the climax criterion, designating the sequence of intervals as an event of the query;
when two designated events satisfy a merge events criterion, merging the two designated events into a single event;
when an interval, between two intervals of the designated event satisfy the climax criterion, satisfies a split event criterion, splitting the designated event into two events; and
storing an indication that the sequence of intervals has been designated as an event.
2. The method of claim 1 wherein length of an event is defined by a start interval and an end interval of the identified sequence, the interval of the frequency that satisfies the climax criterion, and the frequency that satisfies the climax criterion.
3. The method of claim 2 including indicating importance of an event based on the area of the event.
4. The method of claim 3 wherein the area of the event is approximated based on the difference between the start interval and the end interval times the frequency that satisfies the climax criterion.
5. The method of claim 1 wherein the baseline criterion is satisfied when each frequency in the sequence of intervals is above the baseline threshold frequency and wherein the climax criterion is satisfied when a frequency in the sequence of intervals is above the climax threshold frequency and is a locally maximum frequency.
6. The method of claim 1 wherein the split event criterion is satisfied when a frequency at an interval between the intervals of the two climaxes is below a split event threshold.
7. A computer-readable storage medium encoded with instructions for controlling a computing device to identify events of a query, by a method comprising:
identifying sequences of intervals of the query in which the frequency of the query is above a baseline threshold frequency, in which an interval with a maximum frequency of the sequence is above a climax threshold frequency, and in which no interval between the interval with the maximum frequency and closest intervals with frequencies above the climax threshold frequency have a frequency that is below a split event threshold, the frequency for an interval representing number of queries associated with that interval;
designating each identified sequence as an event of the query;
when two events satisfy a merge events criterion, merging the two events into a single event; and
for at least one event,
identifying items of search results of queries submitted during the event; and
designating the items as an explanation for the event.
8. The computer-readable storage medium of claim 7 wherein length of an event is defined by a start interval and an end interval of an identified sequence, the interval of the frequency that satisfies the climax criterion, and the frequency that satisfies the climax criterion.
9. The computer-readable storage medium of claim 8 wherein an area of the event is approximated based on the difference between the start interval and the end interval times the maximum frequency of the sequence of intervals.
10. The computer-readable storage medium of claim 7 wherein the identified items are items selected by users who submitted the query.
11. The computer-readable storage medium of claim 10 including, for the event, generating a list of URLs of items of search results clicked through by users who submitted the query during the event.
12. The computer-readable storage medium of claim 11 wherein the URLs are sorted based on number of clickthroughs.
13. A computing device for identifying events of a query, comprising:
a frequency store storing frequency of the query at intervals, the stored frequency for an interval representing number of queries submitted during that interval;
a memory storing computer-executable instructions of
a component that identifies as candidate events sequences of intervals in which the frequency satisfies a baseline criterion with an interval that satisfies a climax criterion;
a component that merges candidate events when the sequence of intervals satisfies a merge events criterion;
a component that splits candidate events when an interval between two intervals that satisfy the climax criterion satisfies a split event criterion; and
a component that provides an explanation of an event based on selection of search results of the query; and
a processor for executing the computer-executable instructions stored in the memory
wherein a query is represented by a sequence of its events.