METHOD AND DEVICE FOR AUTOMATIC ANNOTATION OF CONTENT OF ELECTRONIC DOCUMENTS

摘要

1. A method of annotating an electronic document, comprising the steps of: splitting an electronic document into a plurality of members, each of the plurality of members being associated with a corresponding length, a corresponding information score and a matching consistency score; a subset of the plurality of members is automatically selected so that the total the informativeness score of this subset is maximum, while the total length of this subset is less than or equal to the maximum length; and include the said subset as an annotation to the electronic document. 2. The method of claim 1, wherein said subset contains less than all of the plurality of members. The method of claim 1, wherein at least one of the members is a proposal. The method of claim 1, wherein the corresponding information score for a given member of the plurality of members is assigned in accordance with a scoring technique that is language independent. The method of claim 4, wherein the scoring technique assigns weights to a plurality of attributes of a given member in accordance with a set of manually programmable rules. The method of claim 1, wherein the corresponding information score for a given member of the plurality of members is assigned in accordance with a scoring technique that is language dependent. The method of claim 6, wherein the scoring technique is a controlled machine learning technique that uses a statistical classifier. The method of claim 7, wherein the statistical classifier corresponds to the support vector method. �