AI &
Patents are a vital tool when it comes to protecting inventions from unauthorized commercialization.
By doing thorough, professional patent searches, inventors can get to know if their inventions are new, useful, and non-obvious. Depending on the findings of a patent search, inventors can then establish beforehand if their inventions are patentable or not.
Need
Inventors have had to contend with the problem of not effectively being able to ascertain the patentability of their inventions. This includes wasting money and time in filing patents for inventions that are not patentable as well as being sued for infringement.
Methodology
We have dedicated our specialized expertise to help inventors leverage AI and automation to perform efficient and exhaustive patent searches for better work outcomes. Apart from the human intelligence in which we believe most of the times, we have been aiding our search process with in-house developed AI based tools and techniques Such tools and techniques are based on combination of open source and in house developed algorithms. In the open source community, algorithms specific to IP domain are less or not available, therefore, algorithms have to be tweaked for applying them in IP domain.
This is how we do it ..
At ResearchWire, we use NLP based techniques, which also includes semantic analysis for quickly uncovering prior arts, reducing costs, and delivering the most accurate prior art findings in a timely manner. We are in the process of building our own worldwide database of patents and are at present using the hybrid methods.
For example, a combination of most relevant keywords and technological classes, according to disclosure/claims of the invention, is used to download a dataset. This is what we call creating a cluster of the relevant documents. NLP based algorithms, which include the use of synonyms from our synonym database, are then applied to the extracted dataset.These algorithms compare the disclosure/claims of the invention, with the full text details of the patents/applications in the extracted dataset based on multiple parameters such as linguistic comparison, technological classes, key concepts, etc.
The patents/applications from the dataset, whose comparison score surpasses a set threshold value, are considered as relevant prior arts. Such prior arts are analyzed by researchers to validate the effectiveness of the algorithms and to ultimately deliver the most accurate prior art search result. The above mentioned hybrid search methodology helps to save on cost as well as the time needed to conduct the search, all this while maintaining the accuracy and quality of the final deliverables. This creates a win-win situation for both, the client and ResearchWire.
Our Capabilities
Researchwire has in-house software capability for AI, NLP and Machine Learning. Our in-house software aims to help researchers focus on critical analysis by identifying potentially good results quickly
We aim to reduce research hours without compromising on work quality. Our software also helps researchers to identify potential areas to explore and as a result, ensures comprehensive coverage of analysis
Our software incorporates our researcher’s decade long understanding of technologies to categorize data into different technologies.
- Communication Technologies
- Networking
- Medical Devices & Diagnostics
- Computer Networks
- Consumer Electronics
- Manufacturing
- Consumer Goods
- Electrical
- Energy
- Semiconductors
- E-Commerce
- Software
