M.Tech Project at IIT Bombay
Abstract As a result of moving into a globalized technological climate, companies search for innovative innovations and secure intellectual property rights to add to their technological competitiveness. In such environments, R&D finds a patent review prior to a new project to be a necessary prerequisite. In terms of scientific knowledge, patent records are an extensive source of technical and commercial knowledge. Progress, business dynamics and proprietary ownership and thus the study of patents has long been seen as. A valuable vehicle for business environment R&D management and macro-context technoeconomic research.
In addition, due to their relative advantages, patents promote analytical work compared to other products. Indexes with regard to database availability, scale of coverage and richness of information. The strategic importance of patent research has recently been further stressed. As the process of innovation becomes more complicated in high-technology management, the duration of Innovation is shortening, and demand from the consumer is becoming more unpredictable.
The purpose of this project is to develop modules / libraries and deploy it later that can perform landscape analysis of any given field or area of study. As a part of phase 1, developed and presented modules that can extract patent data from XML file and store it in CSV format, the analysis of the XML tree was very complex and time consuming also presented and perform bibliographic analysis and contextual analysis of obtained CSV file using Natural Language Processing Techniques.
Bibliographic analysis provides an eagle view of particular patent landscape with which assignee with highest number of patents in a particular time period, which inventor is major contributor in particular field and on which topic is he actually working. Who is collaborating with whom to carry out certain task, etc. But Bibliographic analysis cannot capture the semantics that are captured in patent documents in unstructured format. This is available in Claims, abstract, title etc. and using this unstructured data and performing text analytics methods like Topic modelling with underlying algorithm of LDA (Latent Dirichlet Allocation), gives us cluster of similar patents with taxonomy describing that topic and also provides technology Hotspots and Technology Vacuums and also helps us detect innovative patents using anomaly detection techniques or visual techniques.