5. Summary and future work
In this AZD 6244 article, we presented advanced techniques for patent analysis and summarization that form part of the TOPAS workbench. Each of these techniques addresses a need of IP professionals and constitutes a contribution to the state of the art in the field of patent document processing. Their use increases the efficiency and accuracy of nearly all common applications in the patent domain [3], including analysis of a single patent or a set of patents, landscaping of a patent set, and patent search. In particular, single patent analysis benefits from TOPAS technologies – for instance, for more efficient navigation in a patent or for identification of specific entities (e.g., substances, processes, and measurements) and their distribution in the patent. Figure 3 shows a pdf-interface that supports entity type-specific visualization in different sections of the document.
Fig. 3. Access of TOPAS technologies via an interactive pdf-interface; provided by Software Brügmann.Figure optionsDownload full-size imageDownload as PowerPoint slide
Different entity types are highlighted in different colors. A click on a specific entity provides more information on
this entity.Figure 4 shows the extraction of measurement entities in a different environment.
Fig. 4. Extraction of measurement units in TOPAS; provided by IntelliSemantic.Figure optionsDownload full-size imageDownload as PowerPoint slide
Fig. 5 and Fig. 6 illustrate the use of TOPAS technologies beyond a single patent analysis. Thus, Figure 5 shows their use for correlating different entities to individual patents in a given set.
Fig. 5. Patent – entity correlation visualization using TOPAS technologies and Matheo Analyzer Sotfware; provided by IALE.Figure optionsDownload full-size imageDownload as PowerPoint slide
Fig. 6. Search results selected by specific segments; here: Background Art; provided by IntelliSemantic.Figure optionsDownload full-size imageDownload as PowerPoint slide
Figure 6 exemplifies the exploitation of segmentation for the restriction of the results of a search to specific segments such as, e.g., Background Art.
The TOPAS techniques are integrated into GATE and organized as a pipeline, which can be run in batch mode or integrated in an interactive shell, depending on the needs of the users. The fact that the GATE environment is freely available and open source further facilitates their exploitation.
However, it should be clear that during the lifetime of TOPAS, the TOPAS technologies have been tested on a specific (although representative) patent collection, WIPO\'s IPC Green Inventory.14 That is, to release TOPAS for off-the-shelf use by a broad range of users, an extensive evaluation campaign on a broad range of patent domains and a subsequent revision and improvement of the techniques would be necessary first. The suboptimal performance of the preprocessing tasks (e.g., as tagging, chunking and dependency parsing) furthermore conditions the quality of the techniques. As PATExpert [30], which can be considered its predecessor in our effort to develop advanced intelligent techniques for patent processing applications, TOPAS is thus a prototypical implementation whose technologies require further maturation. This applies to both more restricted technologies such as named entity recognition and broader technologies such as summarization – although the more restricted technologies are by their nature closer to the mature state than others. Thus, it will take much less effort to deploy entity recognition or segmentation than summarization. In any case, the maturation of all technologies will be subject of our future work – as has already been foreseen in TOPAS\' work plan. This also includes the perfection of the preprocessing tasks. Still, in general, it can be stated that all TOPAS techniques are considered to have achieved TRL15 6 at the end of the proje