Relevant individual searching using managed property and ranking features

• US 8,676,784 B2
• Filed: 07/09/2012
• Issued: 03/18/2014
• Est. Priority Date: 10/29/2009
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Li, Boxin, Meyerzon, Dmitriy, Alspaugh, Jessica, Poznanski, Victor
• Primary Examiner(s): Ortiz Ditren, Belix M
• Application Number:
  US13/544,722
• Publication Number:
  US 20120278304A1
• Time in Patent Office: 617 Days
• Field of Search: 707/711
• US Class Current: 707/711
• CPC Class Codes: G06F 16/335 Filtering based on additional data, e.g. user or group profiles filtering in web context G06F16/9535, G06F16/9536

Abstract

Embodiments are configured to provide information relevant to individuals of interest to a searching user. In an embodiment, a method includes identifying relevant individuals of a network using a relevance model that includes the use of a number of managed properties and ranking features to identify relevant individuals of a defined network. The relevance model of one embodiment is defined by a schema that includes a textual matching ranking feature, social distance ranking feature, a levels to top ranking feature, and a proximity ranking feature.

Claims

1. A search engine operating with a processor and configured with:

a first relevancy determination stage that includes a textual matching ranking feature, a social distance ranking feature, and a levels to the top (LTT) ranking feature to identify a number of preliminary search results, wherein the LTT ranking feature is used to generate a weight adjusting multiplier that uses a transformation operation comprising an inverse rational transformation to generate the weight adjusting multiplier; and

,

a second relevancy determination stage that includes a proximity ranking feature that operates on the number of preliminary search results, wherein the proximity ranking feature is used to determine weights to adjust relevancy scores of the preliminary search results and the search engine is further configured during the second relevancy determination stage to identify relevant individuals based in part on the inverse rational transformation operation associated with the LTT ranking feature and a transformation operation associated with the proximity ranking feature.

2. The search engine of claim 1 coupled to at least one server computer and configured to use the proximity ranking feature only as part of the second relevancy determination stage.

3. The search engine of claim 1 further configured to use a schema and a number of managed properties that include one or more of a contents property, an account name property, a preferred name property, a work email property, a username property, a contents hidden property, and a combined name property.

4. The search engine of claim 1 further configured to assign weights to query terms based in part on a ranking equation and one or more managed properties and associated query terms.

5. The search engine of claim 1 further configured to use a relevance model to identify one or more relevant individuals of interest based in part on received query terms and an identity or other identifying parameter to identify a searching user.

6. The search engine of claim 1 further configured to return documents and other information that correspond to relevant people that are associated with one or more different teams, groups, and other associations.

7. The search engine of claim 1 further configured to use the social distance ranking feature in part to adjust ranking weights of relevant individuals based in part on a determination of colleagues associated with a searching user.

8. The search engine of claim 1 further configured to use the social distance ranking feature to adjust ranking weights of relevant individuals including using a first applied weight for a first level colleague and a second applied weight for a second level colleague.

9. The search engine of claim 1 further configured to search for information associated with a plurality of networked users based in part on a plurality of managed property hits.

10. The search engine of claim 1 further configured to use a schema-defined relevancy model including one or more weighting, constant, and statistical parameters.

11. A server configured to identify individuals of interest and configured with:

a first stage of a relevancy model that includes a textual matching ranking parameter, a social distance ranking parameter, and a levels to the top (LTT) ranking parameter to identify preliminary search results, wherein the LTT ranking parameter is used to generate a weight adjusting multiplier including using a first transformation operation comprising an inverse rational transformation to generate the weight adjusting multiplier; and

,

a second stage of the relevancy model that uses a proximity ranking parameter in part to refine the preliminary search results, wherein the proximity ranking parameter is used to determine weights to adjust relevancy scores of the preliminary search results and the relevancy model further includes a second transformation to transform a proximity value in part to provide relevant individual information.

12. The server of claim 11 further configured to use a managed property weight, a property identifier, standard deviation and mean parameters as part of returning relevant individual search results.

13. The server of claim 11 further configured to use a function to enable the proximity ranking feature as part of identifying the relevant individual search results.

14. A method of using a relevance model to provide relevant search results associated with individuals of interest that comprises:

using a textual matching ranking feature during a first relevancy stage to determine query term weights using ranking and managed property features associated with query terms;

using a social distance ranking feature during the first relevancy stage to boost ranking weights of relevant search results;

using a LTT ranking feature during the first relevancy stage to generate a weight adjusting multiplier including using an inverse rational transformation to generate the weight adjusting multiplier; and

,

using a proximity ranking feature during a second relevancy stage to refine the relevant results from the first relevancy stage including determining weights to adjust relevancy scores of the relevant results from the first relevancy stage based in part on query term hits associated with a set of managed properties.

15. The method of claim 14 further configured to use the second relevancy stage to resort a number of relevant results generated by the first relevancy stage.

16. The method of claim 14 further configured to use the textual matching ranking feature to determine one or more weights for query terms of a received query using an equation and a plurality of managed property parameters for each query term hit for each managed property of each identified individual.

17. The method of claim 14 further configured to use the social distance ranking feature to add a first weight to a score of each relevant individual of a first grouping, add a second weight to a score of each relevant individual of a second grouping, and add no value to a score of each relevant individual of a third grouping.

18. The method of claim 14 further configured to infer social distance information based in part on email and other electronic communications.

19. The method of claim 18 further configured to present inferred social distance information for approval, editing, and rejecting.