Understanding and customizing stopword lists for enhanced patent mapping

While the use of patent mapping tools is growing, the ‘black-box’ systems involved do not generally allow the user to interfere further than the preliminary retrieval of documents. Except, that is, for one thing: the stopword list, i.e. the list of ‘noise’ words to be ignored, which can be modified to one’s liking and dramatically impacts the final output and analysis. This paper invokes information science and computer science to provide clues for a better understanding of the stopword lists’ origin and purpose, and how they fit in the mapping algorithm. Further, it stresses the need for stopword lists that depend on the document corpus analyzed. Thus, the analyst is invited to add and remove stopwords—or even, in order to avoid inherent biases, to use algorithms that can automatically create ad hoc stopword lists.