Patent
Automated ranking of contributors to a knowledge base
| العنوان: | Automated ranking of contributors to a knowledge base |
|---|---|
| Patent Number: | 9,594,756 |
| تاريخ النشر: | March 14, 2017 |
| Appl. No: | 13/841830 |
| Application Filed: | March 15, 2013 |
| مستخلص: | A system and method is provided to rank contributors to a knowledge base. In an automated operation, a ranking value is calculated for each of a plurality of knowledge-based contributors based both on document information and on personal network information pertaining to the respective contributor. The document information identifies relationships being documents in the knowledge base and the particular contributor, indicating, for example, whether the contributor authored the document, read the document, or rated the document. The personal network information indicates personal connections in the knowledge base between the respective contributors. |
| Inventors: | Sabharwal, Navin (New Delhi, IN) |
| Assignees: | HCL America Inc. (Sunnyvale, CA, US) |
| Claim: | 1. A method comprising: accessing one or more memories that store document information about documents in a knowledge base, the document information identifying relationships between the documents and contributors to the knowledge base, and personal network information with respect to personal networks of respective contributors to the knowledge base, each personal network defining personal connections between respective contributors forming part of the personal network, wherein each personal network is structured such that nodes of the network are provided by the respective contributors, each personal connection comprises a connection defined directly between two of the nodes of the personal network; and in an automated operation using one or more processors, calculating respective ranking values for two or more of the contributors, the calculating of the respective ranking values comprising, for each of the two or more contributors: calculating a document factor calculated as a quantified value based on one or more properties of the document information for the corresponding contributor, calculating a personal network factor calculated as a quantified value based on one or more properties of the personal network information of the corresponding contributor, and calculating the ranking actor based at least in part on the document factor and based at least in part on the personal network factor. |
| Claim: | 2. The method of claim 1 , wherein: the document factor represents a quantified cumulative value of contributions of the contributor to documents in the knowledge base; and the personal network factor represents a quantified value of the defined relationships between the contributor and other contributors in the knowledge base. |
| Claim: | 3. The method of claim 2 , wherein the calculating of the document factor for a particular contributor comprises calculating respective document values for multiple documents to which the particular contributor has contributed. |
| Claim: | 4. The method of claim 3 , wherein the calculating of the document value for a particular document is based at least in part on: a readership volume parameter representative of the number of contributors who have read the particular document; and a ratings parameter derived from ratings of the particular document by respective contributors. |
| Claim: | 5. The method of claim 4 , wherein the calculating of the document value comprises a weighted combination of the ratings parameter and the readership volume parameter, the ratings parameter being given a greater weight than the readership volume parameter. |
| Claim: | 6. The method of claim 3 , wherein calculating the document factor further comprises: determining a particular contribution type for each of the multiple documents, the contribution type being selected from a predefined set of contribution types indicative of the nature of respective relationships between the particular contributor and the relevant document; applying a weighting to each document value based on the respective determined contribution type, to obtain a weighted document value for each of the multiple documents; and summing the multiple weighted document values. |
| Claim: | 7. The method of claim 6 , wherein the set of contribution types consists of being an author of the document, being a rater of the document, and being a reader of the document, the applying of weightings to the documents comprising giving the greatest weighting to documents for which the particular contributor is the author. |
| Claim: | 8. The method of claim 7 , wherein a ratio of the weighting for being the author relative to the weighting for being a rater is in the range 1:0.55-1:0.65. |
| Claim: | 9. The method of claim 2 , wherein the personal network information indicate structured follower networks associated with the knowledge base, each structured follower network comprising one or more followers who are registered to follow a target contributor. |
| Claim: | 10. The method of claim 9 , wherein the calculating of the personal network factor for a particular contributor is based at least in part on respective ranking values of the one or more followers in the follower network of which the particular contributor is the target contributor. |
| Claim: | 11. The method of claim 10 , wherein the calculating of the personal network factor comprises summing the ranking values of the one or more followers in the follower network of the particular contributor. |
| Claim: | 12. The method of claim 10 , wherein the calculating of the ranking values for the two or more contributors is an iterative process comprising repeated calculation of interim ranking values for the two or more contributors. |
| Claim: | 13. The method of claim 12 , wherein the iterative process comprises calculating the interim ranking values in a sequence determined at least in part by respective initial ranking values, with lower initial ranking values being earlier in the sequence. |
| Claim: | 14. The method of claim 13 , wherein, after a first iteration of the iterative process, the initial ranking values comprise the interim ranking values produced by an immediately preceding iteration of the iterative process. |
| Claim: | 15. The method of claim 13 , further comprising determining the sequence based at least in part on a document volume parameter that indicates a quantity of documents to which respective contributors contributed, the calculating of the ranking values further comprising: determining that a plurality of the contributors have equal initial ranking values; and forming the sequence such that, of the plurality of contributors, those with higher document volume parameters are later in the sequence. |
| Claim: | 16. The method of claim 1 , further comprising: receiving a search query including one or more search parameters; identifying candidate contributors in the knowledge base that satisfy the search parameters, the determining of the ranking values being performed responsive to the search query; and producing a search result based at least in part on the determined ranking values. |
| Claim: | 17. The method of claim 16 , wherein the search query indicates one or more targeted knowledge topics selected from a predefined set of knowledge topics, and wherein the determining of ranking values further comprises filtering the knowledge base to exclude from information on which the determining is based documents and/or contributors who are not linked by associated metadata to the targeted knowledge topics. |
| Claim: | 18. A system comprising: a knowledge base information module comprising one or more computer processor devices configured to access one or more memories that store document information about documents in a knowledge base, the document information identifying relationships between the documents and contributors to the knowledge base, and personal network information with respect to personal networks of respective contributors to the knowledge base, each personal network defining personal connections between respective contributors forming part of the personal network, wherein each personal network is structured such that nodes of the network are provided by the respective contributors, each personal connection comprises a connection defined directly between two of the nodes of the personal network; a ranking calculator comprising at least one computer processor device configured to perform one or more automated operations to calculate respective ranking values for two or more of the contributors, the ranking calculator comprising: a document factor module to calculate for each of the two or more contributors a document factor calculated as a quantified value based on one or more properties of the document information for the corresponding contributor, and a personal network factor module to calculate for each of the two or more contributors a personal network factor calculated as a quantified value based on one or more properties of the personal network information of the corresponding contributor, the ranking calculator being configured to calculate the respective ranking factors based at least in part on the corresponding document factor and based at least in part on the corresponding personal network factor. |
| Claim: | 19. The system of claim 18 , wherein the document factor module is configured to calculate for each of the two or more contributors a document factor that represents a quantified cumulative value of contributions of the contributor to documents in the knowledge base. |
| Claim: | 20. The system of claim 19 , wherein the document factor module comprises a document value module to calculate respective document values for multiple documents to which the particular contributor has contributed. |
| Claim: | 21. The system of claim 20 , wherein the document value module is configured to calculate the document value for a particular document based at least in part on: a readership volume parameter representative of the number of contributors who have read the particular document; and a ratings parameter derived from ratings of the particular document by respective contributors. |
| Claim: | 22. The system of claim 21 , the document value module is configure to calculate the document value in an operation that comprises weighted combination of the ratings parameter and the readership volume parameter, the ratings parameter being given a greater weight than the readership volume parameter. |
| Claim: | 23. The system of claim 20 , wherein the document factor module is further configured to calculate the document factor by a process that comprises: determining a particular contribution type for each of the multiple documents, the contribution type being selected from a predefined set of contribution type indicative of the nature of respective relationships between the particular contributor and the relevant document; applying a weighting to each document value based on the respective determined contribution type, to obtain a weighted document value for each of the multiple documents; and summing the multiple weighted document. |
| Claim: | 24. The system of claim 23 , wherein the set of contribution types consists of being an author of the document, being a rater of the document, and being a reader of the document, the applying of weightings to the documents comprising giving the greatest weighting to documents for which the particular contributor is the author. |
| Claim: | 25. The system of claim 24 , wherein a ratio of the weighting for being the author relative to the weighting for being a rater is in the range 1:0.55-1:0.65. |
| Claim: | 26. The system of claim 19 , further comprising a contributor network module to maintain follower networks that are indicated by the personal network information, each structured follower network comprising one or more followers who are registered to follow a target contributor. |
| Claim: | 27. The system of claim 26 , wherein the personal network factor module is configured to calculate the personal network factor for a particular contributor based at least in part on respective ranking values of the one or more followers in the follower network of which the particular contributor is the target contributor. |
| Claim: | 28. The system of claim 27 , wherein the personal network factor module is configured to calculate the personal network factor by summing the ranking values of the one or more followers in the follower network of the particular contributor. |
| Claim: | 29. The system of claim 27 , wherein the ranking calculator is configured to calculate the ranking values for the two or more contributors in an iterative process comprising repeated calculation of interim ranking values for the two or more contributors. |
| Claim: | 30. The system of claim 29 , wherein the ranking calculator is configured to calculate the interim ranking values in an iterative process that comprises calculating the interim ranking values in a calculation sequence, the system further comprising a sequencing module to determine the calculation sequence based at least in part on respective initial ranking values, with lower initial ranking values being earlier in the calculation sequence. |
| Claim: | 31. The system of claim 30 , wherein the sequencing module is configured such that, after a first iteration of the iterative process, the initial ranking values comprise the interim ranking values produced by an immediately preceding iteration of the iterative process. |
| Claim: | 32. The system of claim 30 , wherein the sequencing module is further configured to determine the calculation sequence based at least in part on a document volume parameter that indicates a quantity of documents to which respective contributors contributed, the ranking calculator further being configure to calculate the ranking values in an operation comprising: determining that a plurality of the contributors have equal initial ranking values; and forming the sequence such that, of the plurality of contributors, those with higher document volume parameters are later in the sequence. |
| Claim: | 33. The system of claim 18 , further comprising a search module to: receive a search query including one or more search parameters; identify candidate contributors in the knowledge base that satisfy the search parameters; and produce a search result based at least in part on the ranking values for the respective candidate contributors. |
| Claim: | 34. The system of claim 33 , wherein the search query indicates one or more targeted knowledge topics selected from a predefined set of knowledge topics, and wherein search module is configured to filter the knowledge base to exclude from information on which calculation of the ranking values is based documents and/or contributors who are not linked by associated metadata to the targeted knowledge topics. |
| Claim: | 35. A non-transitory machine-readable storage medium storing instructions which, when performed by a machine, cause the machine to: access one or more memories that store document information about documents in a knowledge base, the document information identifying relationships between the documents and contributors to the knowledge base, and personal network information that with respect to personal networks of respective contributors to the knowledge base, each personal network defining personal connections between respective contributors forming part of the personal network, wherein each personal network is structured such that nodes of the network are provided by the respective contributors, each personal connection comprises a connection defined directly between two of the nodes of the personal network; and calculate respective ranking values for two or more of the contributors, the calculating of the respective ranking values comprising, for each of the two or more contributors: calculating a document factor calculated as a quantified value based on one or more properties of the document information for the corresponding contributor, calculating a personal network factor calculated as a quantified value based on one or more properties of the personal network information of the corresponding contributor, and calculating the ranking factor based at least in part on the document factor and based at least in part on the personal network factor. |
| Patent References Cited: | 6038560 March 2000 Wical 6598046 July 2003 Goldberg et al. 6718324 April 2004 Edlund et al. 7747630 June 2010 Scofield et al. 8170971 May 2012 Wilson et al. 9461876 October 2016 Van Dusen 2002/0120619 August 2002 Marso et al. 2008/0201348 August 2008 Edmonds et al. 2009/0125403 May 2009 Li 2010/0192005 July 2010 Das 2010/0205055 August 2010 Saraswati 2010/0274815 October 2010 Vanasco 2011/0167066 July 2011 Picault et al. 2011/0231363 September 2011 Rathod 2012/0089617 April 2012 Frey 2012/0102048 April 2012 Heer 2012/0158668 June 2012 Tu et al. 2013/0218866 August 2013 Qian et al. 2014/0129371 May 2014 Wilson et al. |
| Other References: | Allahbakhsh, Mohammad, et al., “An Analytic Approach to People Evaluation in Crowdsourcing Systems”, Technical Report UNSW-CSE-TR-201204, The University of New South Wales, Sydney, Australia, Feb. 2012, 23 pages. cited by examiner Ben Jabeur, Lamjed, et al., “A social model for Literature Access: Towards a weighted social network of authors”, RIAO '10, Paris, France, Apr. 28-30, 2010, pp. 32-39. cited by examiner Cabanac, Guillaume, “Accuracy of inter-researcher similarity measures based on topical and social clues”, Scientometrics, vol. 87, © 2011, pp. 597-620. cited by examiner Cazella, Silvio César, et al., “Modeling User's Opinion Relevance to Recommending Research Papers”, UM 2005, LNAI 3538, Springer-Verlag, Berlin, © 2005, pp. 327-331. cited by examiner Deng, Hongbo, et al., “Formal Models for Expert Finding on DBLP Bibliography Data”, 2008 Eight IEEE International Conference on Data Mining, IEEE Computer Society, © 2008, pp. 163-172. cited by examiner Harris, Christopher, et al., “Using Hybrid Methods for Relevance Assessments in TREC Crowd '12”, Iowa Univ., Iowa City, IA, © 2012, 10 pages. cited by examiner Kamps, Jaap, “The Impact of Author Ranking in a Library Catalogue”, BooksOnline '11, Glasgow, Scotland, UK, Oct. 24, 2011, pp. 35-39. cited by examiner Kazai, Gabriella, et al., “Overview of the INEX 2010 Book Track: Scaling Up the Evaluation Using Crowdsourcing”, INEX 2010, LNCS 6932, Springer-Verlag, Berlin, © 2011, pp. 98-117. cited by examiner Middleton, Stuart E., et al., “Exploiting Synergy Between Ontologies and Recommender Systems”, Semantic Web Workshop 2002, Hawaii, May 2002, pp. 1-10. cited by examiner Riggs, Tracy, et al., “An Algorithm for Automated Rating of Reviewers”, JCDL '01, Roanoke, VA, Jun. 24-28, 2001, pp. 381-387. cited by examiner Rosen-Zvi, Michel, et al., “The Author-Topic Model for Authors and Documents”, UAI '04, Arlington, VA, © 2004, pp. 487-494. cited by examiner Tamine, Lynda, et al., “Chapter 6: On Using Social Context to Model Information Retrieval and Collaboration in Scientific Research Community”, Community-Built Databases, Springer-Verlag, Berlin, © 2011, pp. 133-155. cited by examiner Volkovs, Maksims N., et al., “Learning to Rank by Aggregating Expert Preferences”, CIKM '12, Maui, HI, Oct. 29-Nov. 2, 2012, pp. 843-851. cited by examiner Yang, Kai-Hsiang, et al., “Author Name Disambiguation for Citations Using Topic and Web Correlation”, ECDL 2008, Springer-Verlag, Berlin, © 2008, pp. 185-196. cited by examiner Datta, Anwitaman, “SoJa: Collaborative Reference Management Using a Decentralized Social Information System”, CollaborateCom 2010—Session 3: Online Social Networks, Chicago, IL, Oct. 9-10, 2010, pp. 1-9. cited by examiner Chen, Chaomei, et al., “Visualizing a Knowledge Domain's Intellectual Structure”, Computer, vol. 34, Issue 3, Mar. 2001, pp. 65-71. cited by examiner |
| Primary Examiner: | Stevens, Robert |
| Attorney, Agent or Firm: | Schwegman Lundberg & Woessner, P.A. |
| رقم الانضمام: | edspgr.09594756 |
| قاعدة البيانات: | USPTO Patent Grants |
| الوصف غير متاح. |