Method and system for disambiguating informational objects
The present invention provides a Distinct Author Identification System (“DAIS”) for disambiguating data to discern author entities and link or associate authorships with such author entities. The invention provides powerful disambiguation processes applied across one or more databases to yield a disambiguated authority database of authors. An entire database of publications may be processed by the DAIS to group/link authorships and to identify author entities. The author entities may then be matched or associated with actual authors to establish an authority database of authors. After initial evaluation, the DAIS may be used to reevaluate some or all of the database(s) and/or the authority database established by the DAIS may be used to add or update information. DAIS may use “hierarchical clustering” to link authorships and identify authors based on authorship similarity. DAIS evaluates the likelihood that authorships are from the same author.
1. A computer implemented method for maintaining an authority database comprising a plurality of unique author profile records associated with authors of publications, the method comprising
receiving data representing a user-defined set of publications, each publication having at least one authorship and related to a unique author of the publication;
recognizing the received data as being associated with a researcher identifier;
by using the authority database comprising the plurality of unique author profile records, verifying the received data to render a threshold confirmation of correctness in association of the set of publications with the unique author, wherein each unique author profile record is associated with a unique author identifier and a cluster of authorships associated with the unique author identifier; and
matching the unique author with an existing unique author profile record stored by the authority database, or creating a new unique author profile record and storing the new unique author profile record at the authority database, wherein
the existing unique author profile record includes a cluster of a set of authorships of a set of publications attributed to a unique author represented by the existing unique author profile record.
2. The method of claim 1 , wherein the unique author profile record resulted from:
receiving publications, each publication containing at least one cited reference and having at least one authorship; and
disambiguating the received publications by comparing the at least one cited reference with data associated with the authority database of authors to determine an authorship similarity between publication authorships.
3. The method of claim 2 , wherein disambiguating includes:
scoring an authorship similarity; and
linking authorships based on the determined authorship similarity and clustering two or more linked authorships to form a first cluster and forming a first author entity associated with the first cluster.
4. The method of claim 3 , wherein disambiguating includes:
matching the first author entity with a first actual author, the first cluster of authorships being attributable to the first actual author, and repeating the clustering step to form a plurality of clusters respectively associated with a plurality of unique author entities; and
incorporating into the authority database of authors the plurality of unique author entities each associated with a unique actual author and a cluster.
5. A computer-based system comprising:
a computer including processing circuitry configured to:
process a set of electronic information associated with a set of publications, each publication in the set of publications comprising at least one cited reference and having at least one authorship,
receive data representing a user-defined set of publications, each publication having at least one authorship and related to a unique author of the publication,
recognize the received data as being associated with a researcher identifier,
access an authority database and verify the received data to render a threshold confirmation of correctness in association of the set of publications with the unique author, wherein the authority database comprises a plurality of unique author profile records, such that each unique author profile record is associated with a unique author identifier and a cluster of authorships associated with the unique author identifier, and
process the received data to match the unique author with the existing unique author profile records stored by the authority database, or create a new unique author profile record and storing the new unique author profile record at the authority database, wherein
the existing unique author profile record includes a cluster of a set of authorships of a set of publications attributed to a unique author represented by the existing unique author profile record.
6. The system of claim 5 , wherein the unique author profile record resulted from the processing circuitry:
receiving publications, each publication containing at least one cited reference and having at least one authorship; and
disambiguating the received publications by comparing the at least one cited references with data associated with the authority database of authors to determine an authorship similarity between publication authorships.
7. The system of claim 6 , wherein disambiguating includes:
scoring an authorship similarity; and
linking authorships based on the determined authorship similarity and clustering two or more linked authorships to form a first cluster and forming a first author entity associated with the first cluster.
8. The system of claim 7 , wherein disambiguating includes:
matching the first author entity with a first actual author, the first cluster of authorships being attributable to the first actual author, and repeating the clustering step to form a plurality of clusters respectively associated with a plurality of unique author entities; and
incorporating into the authority database of authors the plurality of unique author entities each associated with a unique actual author and a cluster.