Columbia University

International Business Machines Corporation

Judith L Klavans

Kathleen R McKeown

Wlodek Wlodzimierz Zadrozny

Michelle Yoonkyung Kim

Regina Barzilay

Smaranda Muresan

Vasileios Hatzivassiloglou

Min-Yen Kan

Ani Nenkova

Lisa C Braden-Harder

Dave Evans

Barry Schiffman

A system for generating a summary of a plurality of documents and presenting the summary information to a user is provided which includes a computer readable document collection containing a plurality of related documents stored in electronic form. Documents can be pre-processed to group documents into document clusters. The document clusters can also be assigned to predetermined document categories for presentation to a user. A number of multiple document summarization engines are provided which generate summaries for specific classes of multiple documents clusters. A summarizer router is employed to determining a relationship of the documents in a cluster and select one of the document summarization engines for use in generating a summary of the cluster. A single event engine is provided to generate summaries of documents which are closely related temporally and to a specific event. A dissimilarity engine for multiple document summary generation is provided which generates summaries of document clusters having documents with varying degrees of relatedness. A user interface is provided to display categories, cluster titles, summaries, related images.

System and method for document collection, grouping and summarization

A system for automatically generating a dictionary from full text articles extracts &#x3c;term, definition&#x3e; pairs from full text articles and stores the &#x3c;term, definition&#x3e; pairs as dictionary entries. The system includes a computer readable corpus having a plurality of documents therein. A pattern processing module () and a grammar processing module () are provided for extracting &#x3c;term, definition&#x3e; pairs from the corpus and storing the &#x3c;term, definition&#x3e; pairs in a dictionary database (). A routing processing module selectively routes sentences in the corpus to at least one of the pattern processing module or grammar processing module. In one embodiment, the routing module is incorporated into the pattern processing module which then selectively routes a portion of the sentences to the grammar processing module. A bootstrapping processing module () can be used to apply &#x3c;term, definition&#x3e; entries against the corpus to identify and extract additional &#x3c;terms, definition&#x3e; entries.

System and method of generating dictionary entries

A &ldquo;domain-general&rdquo; method for topical segmentation of a document input includes the steps of: extracting one or more selected terms from a document; linking occurrences of the extracted terms based upon the proximity of similar terms; and assigning weighted scores to paragraphs of the document input corresponding to the linked occurrences. In accordance with the present invention, the values of the assigned scores depend upon the type of the selected terms, e.g., common noun, proper noun, pronominal, and the position of the linked occurrences with respect to the paragraphs, e.g., front, during, rear, etc. Upon zero-sum normalization, the assigned scores represent the boundaries of the topical segments of the document input.

Method and system for topical segmentation, segment significance and segment function

To archive information, a phrase or sentence describing the information, typically expressed in a natural language and conforming to the rules of a grammar (like a natural language grammar) is used to create a structured index which also conforms to the natural language grammar. The structured index has structure because the words in the index have a function and a relationship among each other as determined by the grammar. The index is combined with a location pointer of information to be cataloged, preferably multimedia information, to form a matched pair, i.e., a structured index and a segment (or a pointed to a multimedia object). The matched pair is stored in a database for later retrieval. Retrieval is based on parsing a query into a query structure that is identical or similar to the structured index. A heuristic interface presents the multimedia information along with a template in order to prompt the user into describing the multimedia information with a string of words so that they conform to the grammar.

Archiving and retrieving multimedia objects using structured indexes

Growing community of inventors

Hastings-on-Hudson, NY, United States of America

Judith L Klavans