Entity resolution from documents

1. A method for resolving entities from a plurality of documents, the method comprising:

obtaining, by a processor, the plurality of documents corresponding to a plurality of entities, from at least one data source;

blocking, by the processor, the plurality of documents into one or more buckets based on textual similarity and inter-document references among the plurality of documents, wherein the inter-document references are indicative of relations in the plurality of documents, wherein the inter-document references are identified through corresponding referential attributes, wherein the referential attributes are selected from one of a hard referential attribute and a unique referential attribute of a referred document, wherein the hard referential attribute is an attribute for which two entries are considered to be matching only if the two entries are textually identical and the unique referential attribute is an attribute that has a unique entry for each entity in the plurality of documents, wherein the referential attributes are one of an implicit referential attribute and an explicit referential attribute, wherein the implicit referential attribute comprises a part of its value same as one of the hard referential attribute and the unique referential attribute of the referred document, wherein the explicit referential attribute is an exact value of one of the hard referential attribute and the unique referential attribute of the referred document, and wherein a Locality Sensitive Hashing (LSH) technique in combination with a document traversal technique is used to block the plurality of documents into the one or more buckets;

performing a pair-wise comparison of the plurality of documents in each bucket to identify a set of documents relating to an entity by using an iterative match-merge technique (IMM) based on a pre-defined criteria, wherein the iterative match-merge technique identifies, from the plurality of documents, at least one matching pair of documents and merges the at least one matching pair of documents to create at least one merged documents for each entity in the one or more buckets, wherein the at least one merged document for each entity from the one or more buckets is referred to as partial entities from the one or more buckets;

connecting the partial entities from the one or more buckets sharing the same document to generate a node-edge structure for each of the partial entities from the one or more buckets, wherein each partial entity has a number of associated documents, and each document is represented by a node in the node-edge structure, wherein all nodes belong to two partial entities sharing the same document are connected, and considered to be belonging to the same entity; and

merging all the connected documents representing connected nodes from node-edge structures of the partial entities, belonging to the same entity, from the one or more buckets to form a resolved entity-document for each entity.

2. The method as claimed in claim 1 further comprising updating a resolved entity-document collection upon receiving a new set of documents, wherein an entity resolution analysis is continued without being restarted when the new set of documents are received, and the new set of documents are accommodated in the entity resolution analysis based on the entity resolution analysis already performed, wherein the updating is performed based on the textual similarity and the inter-document references among the new set of documents and the resolved entity-documents.

3. The method as claimed in claim 1 , wherein the at least one matching pair of documents is identified based on the textual similarity and the inter-document references among the plurality of documents.

4. The method as claimed in claim 1 , wherein the merged document for each entity is created using an R-Swoosh technique.

5. An entity resolution system for entity resolution from a plurality of documents, the entity resolution system comprising:

a processor;

a blocking module, coupled to the processor, to,

obtain the plurality of documents corresponding to a plurality of entities, from at least one data source; and

block the plurality of documents into one or more buckets based on textual similarity and inter-document references among the plurality of documents, wherein the inter-document references are indicative of relations in the plurality of documents, wherein the inter-document references are identified through corresponding referential attributes, wherein the referential attributes are selected from one of a hard referential attribute and a unique referential attribute of a referred document, wherein the hard referential attribute is an attribute for which two entries are considered to be matching only if the two entries are textually identical and the unique referential attribute is an attribute that has a unique entry for each entity in the plurality of documents, wherein the referential attributes are one of an implicit referential attribute and an explicit referential attribute, wherein the implicit referential attribute comprises a part of its value same as one of the hard referential attribute and the unique referential attribute of the referred document, wherein the explicit referential attribute is an exact value of one of the hard referential attribute and the unique referential attribute of the referred document, and wherein a Locality Sensitive Hashing (LSH) technique in combination with a document traversal technique is used to block the plurality of documents into the one or more buckets; and

a merging module, coupled to the processor, to,

perform a pair-wise comparison of the plurality of documents in each bucket to identify a set of documents relating to an entity by using an iterative match-merge technique (IMM) based on a pre-defined criteria, wherein the iterative match-merge technique identifies, from the plurality of documents, at least one matching pair of documents and merges the at least one matching pair of documents to create at least one merged document for each entity in the one or more buckets, wherein the at least one merged document for each entity from the one or more buckets is referred to as partial entities from the one or more buckets;

connect the partial entities from the one or more buckets sharing the same document to generate a node-edge structure for each of the partial entities from the one or more buckets, wherein each partial entity has a number of associated documents, and each document is represented by a node in the node-edge structure, wherein all nodes belonging to two partial entities sharing the same document are connected, and considered to be belonging to the same entity; and

merge all the connected documents representing connected nodes from node-edge structures of the partial entities, belonging to the same entity, to form a resolved entity-document for each entity.

6. The entity resolution system as claimed in claim 5 further comprising an updating module, coupled to the processor, to update a resolved entity-document collection upon receiving a new set of documents, wherein the resolution of entities is continued without being restarted when the new set of documents are received, and the new set of documents are accommodated in the resolution of entities based on the resolution of entities already performed, wherein the updating is performed based on the textual similarity and the inter-document references among the new set of documents and the resolved entity-documents.

7. The entity resolution system as claimed in claim 5 , wherein the at least one matching pair of documents is identified based on the textual similarity and the inter-document references among the plurality of documents.

8. The entity resolution system as claimed in claim 5 , wherein the merging module creates the merged document for each entity using an R-Swoosh technique.

9. A non-transitory computer-readable medium having embodied thereon a computer program for executing a method comprising:

obtaining, by a processor, a plurality of documents corresponding to a plurality of entities, from at least one data source;

blocking, by the processor, the plurality of documents into one or more buckets based on textual similarity and inter-document references among the plurality of documents, wherein the inter-document references are indicative of relations in the plurality of documents, wherein the inter-document references are identified through corresponding referential attributes, wherein the referential attributes are selected from one of a hard referential attribute and a unique referential attribute of a referred document, wherein the hard referential attribute is an attribute for which two entries are considered to be matching only if the two entries are textually identical and the unique referential attribute is an attribute that has a unique entry for each entity in the plurality of documents, wherein the referential attributes are one of an implicit referential attribute and an explicit referential attribute, wherein the implicit referential attribute comprises a part of its value same as one of the hard referential attribute and the unique referential attribute of the referred document, wherein the explicit referential attribute is an exact value of one of the hard referential attribute and the unique referential attribute of the referred document, and wherein a Locality Sensitive Hashing (LSH) technique in combination with a document traversal technique is used to block the plurality of documents into the one or more buckets;

performing a pair-wise comparison of the plurality of documents in each bucket to identify a set of documents relating to an entity by using an iterative match-merge technique (IMM) based on a pre-defined criteria, wherein the iterative match-merge technique identifies, from the plurality of documents, at least one matching pair of documents and merges the at least one matching pair of documents to create at least one merged document for each entity in the one or more buckets, wherein the at least one merged document for each entity from the one or more buckets is referred to as partial entities from the one or more buckets;

connecting the partial entities from the one or more buckets sharing the same document to generate a node-edge structure for each of the partial entities from the one or more buckets, wherein each partial entity has a number of associated documents, and each document is represented by a node in the node-edge structure, wherein all nodes belonging to two partial entities sharing the same document are connected, and considered to be belonging to the same entity; and

merging all the connected documents representing connected nodes from node-edge structure of the partial entities, belonging to the same entity, from the one or more buckets to form a resolved entity-document for each entity.

10. The method of claim 1 , wherein the referential attributes comprise soft attributes, wherein each soft attribute is an attribute for which two entries of the plurality of documents are considered to be matching, even if the two entries are not textually identical.

11. The system of claim 5 , wherein the referential attributes comprise soft attributes, wherein each soft attribute is an attribute for which two entries of the plurality of documents are considered to be matching, even if the two entries are not textually identical.

12. The non-transitory computer-readable medium of claim 9 , wherein the referential attributes comprise soft attributes, wherein each soft attribute is an attribute for which two entries of the plurality of documents are considered to be matching, even if the two entries are not textually identical.