Summary
The application relates to the field of data retrieval and indexing, and seeks to provide an improvement over the “bag-of-words” method used in natural language processing where independent items (words) taken from an object (a text document) are represented as an unordered collection thereby losing much of the semantic information present in the original document. The proposed method expresses the text as a group of word pairs, disregarding the order of the word pairs. This grouping in the described bag-of-wordpairs (or bag-of-wordgroups where more than 2 words are grouped) method thus allows more of the semantic information to be modelled.
The examiner had objected that the invention was no more than a program for a computer. The hearing officer found that the amended claims, which were restricted to a method of speech processing, did not fall solely within an excluded category. The application was remitted to the examiner for further substantive examination.