Google Inc.

Shubin Zhao

Jonathan T Betz

Krzysztof Czuba

A query is defined that has an answer formed of terms from electronic documents. A repository having facts is examined to identify attributes corresponding to terms in the query. The electronic documents are examined to find other terms that commonly appear near the query terms. Hypothetical facts representing possible answers to the query are created based on the information identified in the fact repository and the commonly-appearing terms. These hypothetical facts are corroborated using the electronic documents to determine how many documents support each fact. Additionally, contextual clues in the documents are examined to determine whether the hypothetical facts can be expanded to include additional terms. A hypothetical fact that is supported by at least a certain number of documents, and is not contained within another fact with at least the same level of support, is presented as likely correct.

Corroborating facts in electronic documents

A system and method for extracting facts from documents. A fact is extracted from a first document. The attribute and value of the fact extracted from the first document are used as a seed attribute-value pair. A second document containing the seed attribute-value pair is analyzed to determine a contextual pattern used in the second document. The contextual pattern is used to extract other attribute-value pairs from the second document. The extracted attributes and values are stored as facts.

Unsupervised extraction of facts

A system and method for corroborating a set of facts. If the anchor text of the references to a document matches the name of a set of facts, the referenced document is used to corroborate the set of facts. By analyzing the anchor text of the references to the document, the system is capable of determining if a document is relevant to the set of facts. These documents can then be used to corroborate or refute the facts, thereby improving their overall quality.

Anchor text summarization for corroboration

A system, method, and computer program product for learning objects and facts from documents. A source object and a source document are selected and a title pattern and a contextual pattern are identified based on the source object and the source document. A set of documents matching the title pattern and the contextual pattern are selected. For each document in the selected set, a name and one or more facts are identified by applying the title pattern and the contextual pattern to the document. Objects are identified or created based on the identified names and associated with the identified facts.

Learning objects and facts from documents

A system, method, and computer program product for recognizing entity names from a plurality of documents. Embodiments of the methods comprise selecting a selection of documents from a plurality of documents, the selection of documents sharing a common pattern in their titles. The method further comprises determining a name candidate for each document in the selection by applying the common pattern to the title of the document, and matching the name candidates with a collection of entity names (the white list). Responsive to determining a match between the name candidates and the entity names in the white list, the method determines that the name candidates are valid entity names. In one embodiment, the name candidates are added to the white list after being determined to be valid entity names.

Entity name recognition

A method and system of learning, or bootstrapping, facts from semi-structured text is described. Starting with a set of seed facts associated with an object, documents associated with the object are identified. The identified documents are checked to determine if each has at least a first predefined number of seed facts. If a document does have at least a first predefined number of seed facts, a contextual pattern associated with the seed facts is identified and other instances of content in the document matching the contextual pattern are identified. If the document includes at least a second predefined number of the other instances of content matching the contextual pattern, then facts may be extracted from the other instances.

Learning facts from semi-structured text

A system and method identifies a subject for a source document. The system and method identifies a collection of peer documents from the same domain as the source document. For each of the peer documents, a collection of linking documents containing a hyperlink to the peer document is identified. For each of the peer documents, a label is generated by choosing the longest-match anchor text of the linking documents. A pattern between the labels and the titles of the collection of peer documents is deduced. The subject of the source document is identified by applying the pattern to the title of the source document.

Growing community of inventors

Jersey City, NJ, United States of America

Shubin Zhao