Automatic classification of text files using word contexts

摘要

Text files 204 are categorised to improve searching of a text file database. Categorisation is performed by identifying the occurrence of relevant words such as company names without direct specification of any company names. This is done by specifying words or phrases that imply the occurrence of a relevant word and searching the text files for occurrences of these words or phrases. For example the phrase "shares in' will tend immediately to precede a company name, and so a search for this phrase will assist identification of files which should be considered for classification as company information. However, if the relevant word is also detected in other contexts 206 which suggest that the word is not relevant, then the probability of association with the preferred context 205 is reduced 210,211 based on a scoring scheme.