Patent Expert Services
Next generation patent analysis for new insights
Expert knowledge and cutting edge AI applications
Human IP experts combine deep technological background knowledge with patent law expertise to act as patent finders and decision makers, not just as patent searchers. By comparison, a modern artificial intelligence is rarely much more than an uneducated child with enormous computing power and endless data storage. The key to next-generation high-quality knowledge extraction lies in combining the two approaches.
EconSight has created a taxonomy of more than 500 technologies that have been gathered, vetted, and selected by human experts with more than 20 years of patent research experience. EconSight’s proprietary AI algorithms are similarly designed to read, understand, and categorize the 130+ million patent documents at the text level and in their technological context. By combining these two worlds, we can find the missing text fragments or long sought documents that the traditional human searcher may have missed. Huge patent landscapes with more than 100’000 patent families can be structured, clustered and categorized into content-separated segments, fully multilabel and purely text-based. Our 3D cluster analysis calculates dynamics, size and conceptual distance from the main field concept to find and identify the most promising segments in large fields: Fully explicable AI down to the single patent and most likely owner.
Major players trust the EconSight approach
New EconSight Study “Hydrogen Technologies in Transport” for WIPO
Over the next decade, transforming the transportation sector to put it on a Net Zero pathway will require a combination of technological innovation, government and corporate decision-making, and adapted customer behavior. Reducing greenhouse gas emissions by transportation, a sector responsible for almost 24 percent of direct carbon dioxide emissions from fuel combustion is crucial. EconSight’s Report for the WIPO Patent Landscape series provides early observations on patenting activity together with complementary information from online news, press releases and corporate financial reporting in the field of hydrogen fuel cells in transportation.
Customized offerings for IP experts
Patent positioning & dynamic landscape evaluation
We investigate the conceptual distribution of a technology field and measure the size and dynamics of the different thematic segments. We thus find the concepts with high density and size, as well as those with high dynamics. We measure not only the general activity, but also specifically the presence and distribution of competitors or the presence of yet unknown but thematically relevant new players. We use state-of-the-art pre-trained AI models and apply them to the text elements of the patents to separate the conceptual clusters well. For this purpose, we already have 130 million global patents with current legal status in pre-calculated, paragraph-divided text sections, in English or machine-translated, and we do not need to load patents from other sources. The thematic contents are characterized by a detailed analysis of the conceptual cluster center-of-gravity in understandable word representations.
Draft & innovation evaluation and scoring, grant predictor
We examine client drafted concepts, innovations or unpublished patent applications according to several factors, such as Grant Predictor and Grant Probability, Uniqueness or Citation-Score Prediction, and position these drafts in a predefined technology landscape as if it already exists as a patent. No text element is used outside of a secure, local environment, and no text is sent to an external database. On customer request, a variant completely closed from the outside world is available, which provides technology field-specific evaluations of a draft or patent application according to a patent-pending method, in order to be able to decide at an early stage which inventions have higher potential.
AI enhanced patent searches
Based on drafts or patent documents, we offer different intensified searches in our globally covered patent database. In all searches, our trained AI will search available text for the maximum similarity based on state-of-the-art transformer technology and produce the text passage as a result. The intensity levels are primarily based on the target text passage (abstract or full text) as well as the scope of the patents to be searched (with or without time or regional restriction). In all cases, our proprietary algorithms first identify the relevant technology field and then search for relevant texts with high precision. In doing so, our engines leverage the breadth of expert-defined technologies for maximum relevance of results. Optional manual pre- or post-analysis can be added.
Neighborhood analysis, infringer search, risk-analysis
We measure the thematic distance of existing patents to a target company in a time rolling view. In this process, we measure in absolute distances which companies are moving towards or away from the target company in terms of topic and text content. The analysis is based purely on the text of the underlying patents and ignores bibliographic or citation dependencies. The target company or entity is predefined as the center of gravity and it becomes significantly visible which companies (as well as universities, geographic regions, countries or other groupable entities) move towards or away from a target in terms of content and over time.