International Business Machines Corporation

Xueliang Zhao

Xin Xu

Kun Yan Yin

He Li

Xin Tang

Zhong Fang Yuan

Tong Liu

Si Tong Zhao

Teng Sun

Yu Zui Wy You

Frank Feng

In order to perform a semantic search based on a graph database, sets of nodes are selected from a plurality of nodes in a graph database. A set of nodes semantically matches a keyword in a natural language query. At least one target node is identified in the sets of nodes. A path is selected from candidate paths based on similarities between the candidate paths and a plurality of paths in the graph database. A graph query for retrieving information from the graph database is generated based on the selected path and the query target.

Semantic search based on a graph database

An approach is provided for searching multilingual documents. A first classification is determined that includes a first document and other document(s) by minimizing a first distance between a first numerical fixed length vector for the first document and other numerical fixed length vector(s) for other document(s). Based on a query and a natural language detected in the query, a second document is selected. A second stream modeling the second document is encoded as a second numerical fixed length vector. Based on a distance between the first and second numerical fixed length vectors being less than a threshold, the first classification is identified as including the second document. Documents in the first classification are ranked and presented as having content matching the second document's content. At least one of the ranked documents is expressed in a natural language different from the natural language of the second document.

Searching multilingual documents based on document structure extraction

An approach is provided for searching multilingual documents. Structure components are extracted from multilingual documents. Based on the extracted components, the documents are grouped into classifications including respective sets of documents expressed in different respective natural languages. A natural language in a query is detected. One of the documents is selected based on the document having content indicated by the query and the natural language of the document matching the detected natural language. Structure components of the selected document are extracted. Based on the extracted structure components of the selected document, one of the classifications is identified as including the selected document. Other document(s) in the classification are identified and presented as having content that matches the content of the selected document. The natural language(s) of the other document(s) are each different from the natural language of the selected document.

Growing community of inventors

Shanghai, China

Xueliang Zhao