Toevolution

Despite the disruption brought on by the COVID-19 epidemic, an increase in patent applications has been observed. In 2020, there were about 3.3 million patent applications submitted, a 1.6% rise from the previous year. An unprecedented chance to gain practical business insights from this massive patent data is presented. We also have patent mining to make sense of patent data, which is akin to the idea of data mining. Patent mining is the process of extracting certain patents from a sizable database. It comprises deciding which IP assets to buy, sell, licence, or otherwise dispose of in order to increase the portfolio's effectiveness. With business objectives in mind, patent mining may be carried out manually or automatically using cutting-edge techniques.

The idea of patent mining, its necessity, many jobs required, and more will all be covered in this article.

Why is Patent Mining Needed?

1. Optimize Patent Portfolio: For companies that invest in a steady supply of new patents, optimising a patent portfolio is a big challenge. Finding the relevant patents in a sizable patent pool necessitates specific knowledge and experience. A tried-and-true method to overcome this issue is patent mining. Patent portfolio optimization may provide a number of advantages, such as weeding out low-value patents, identifying patents that are pertinent to new technologies, or getting rid of patents that seem to use up all of an organization's resources.

2. Competitor Analysis: Patent mining is more clever and dependable than both new and ancient strategies for competitor analysis. Investigating a rival's patents can yield fascinating details about the strategy, technology, and user-friendly characteristics of your competitors. Additionally, a business might devise a plan that could assist it in staying one step ahead of the competition in the future by studying the patents of its rivals.

3. Recognize New Technology Trends: Over the past ten years or more, the technological landscape has gotten very complex. To stand out in a new market, it is crucial to understand both current and developing technology. Technology road mapping is facilitated by the patent mining system. The method aids in comparing the organization's technical maturity level to that of its technologically advanced rivals. An organisation can also use the data to investigate new technical alliances, technology transfers, or even the purchase of a start-up.

4. Develop a Dynamic Patent Revenue Model: Through patent mining, a company might locate patents that are strategically important to rival market players. As a result, the exercise enables the monetization of outdated patents through their sale or granting of licences to develop a dynamic patent revenue model. Since technology adoption is not always rapid, certain companies in a given industry may find value in patents relating to very old technologies.

5. Conducting a patent infringement study on a comprehensive list of patents is a time-consuming operation and an antiquated method. Patent mining aids in reducing datasets by determining their relevance. Patents are rated and divided into batches once the dataset has been identified. One batch could contain between one hundred and five hundred patent families. Patents categorised as "T1" have broad claim scope, few detectability problems, and a higher chance of infringement. By taking this step, it is made sure that the patents with the highest likelihood of infringement receive the majority of the effort.

It is critical to comprehend how patent mining is used for infringement analysis after going over the fundamental requirements for it.

How Sagacious IP Analyzes Infringement Using Patent Mining

1. Obtaining Useful Datasets

Extraction of the pertinent patent dataset in accordance with customer needs is the first step in the patent mining process. For instance, it can be necessary to locate patents covering a certain industry, field of technology, or business. In some circumstances, it can simply be a superior market product.

A comprehensive list of pertinent classes and keywords is found and is then used to extract a large amount of data. To extract the dataset, assignee- and inventor-based searches are also carried out.

It's also crucial to make sure the right database is being used to pull data for comprehensive, jurisdiction-based coverage.

Quick manual filtering is then carried out to remove the irrelevant data or the false positives, taking into account the size of the extracted dataset (and the client's budget). Before proceeding to the following stage, automated ranking, the process provides a dataset that is considerably more relevant to deal with.

Autonomous Ranking

The next action is to automatically rate the entire dataset after locating the patent dataset. The dataset's patents are here relatively ranked using measurable criteria. These factors include the number of forward and backward citations, the number of claims, the number of big assignees, the number of family members, the length of an independent claim, the length of the prosecution, etc. Additionally, the factor "forward mentioned by X" is taken into account while evaluating the patents when the goal is to find pertinent patents against a specific target firm (let's say, X).

It is crucial to compare patents with all U.S. patents from the same patent classification and issue year in terms of the number of citations obtained, even while citation analysis aids in the identification of high-impact patents because raw citation counts have a tendency to favour earlier patents.

The automatic ranking intends to prioritise a collection of patents that has a higher chance of infringement for investigation and partition the entire list of patents into smaller batches. The activity aids in producing a better result in less time.

3. T-ranking and triage

Patents with higher rankings in the list are given priority for analysis after the automated ranking is finished.

We undertake manual analysis for each batch of patents (batch size could be 100–500 patent families), spending some time quickly going over the title, abstract, and claims. Then, patents are ranked as T1, T2, and T3 based on the claim scope, likelihood of infringement (for instance, generally or against a specific product category), and infringement detectability. In order to determine which set of claims in a patent family is the widest, the claims of the parent patent as well as those of continuation and divisional applications are taken into account. Patents classified as T1 have broad claim scope, minimal detectability problems, and a higher chance of infringement.

4. Extensive Analysis of Infringement

To determine the novelty of the patent at this point, a more thorough examination of the patent specification and file wrapper is conducted. The patent specification also offers hints about the potential areas of application for each patent.

For the purposes of the analysis, we choose the independent claim from each patent that is the widest.

The top companies by revenue or popularity for the relevant market segments or product categories are discovered and examined when the relevant market areas or suitable product categories are determined for a patent. The goal is to pinpoint high-revenue businesses as the infringing parties.

To evaluate the likelihood of infringement for a specific target or firm, we attempt to study documentation (product manuals, datasheets, source code, research papers), as released by the company itself (since it is more credible). If necessary, we also rely on reliable third-party sources to comprehend the features and functionality of the product and to identify any infringement.

We strive to find as much evidence from public documents as we can in situations when reverse engineering is necessary to demonstrate mapping for one or more claim parts, and we advise customers on reverse engineering appropriately.

Claim mapping

We create a claim chart for the designated targets after identifying the infringing targets.

We create charts with various levels of resolution depending on the client's budget and the end goal. If the goal is to just engage in informal, high-level negotiations with a specific defendant in order to reach a settlement outside of court, we will create one or two slider charts in which a clear, high-level mapping of the infringement is displayed. However, in situations when legal action will be taken, we create elaborate or high-resolution charts that break down the claim's elements to the nth level and provide evidence for each claim clause in great detail.

To explain how the patent specification supports the alleged infringement, comparable excerpts from the specification are also provided.

Each claim element additionally has analysis comments that explain how it relates to the relevant product feature.

A ideal method is to assign a specific colour to each claimed characteristic when categorising claim items. For simplicity of claim to product linkage, the corresponding proof is likewise marked with the same colour.

After talking about the need for patent mining and how Sagacious IP employs it for infringement analysis, we'll concentrate on the tasks and methods for mining patents that yield the greatest results.

What are Patent Mining Tasks and Techniques

Data that is both structured and unstructured can be found in patent documents. In structured data, the meanings and formats are consistent (patent number, filing date, inventor, issued date, and assignees). However, unstructured things include text content with varying lengths (such as claims, abstracts, and descriptions of the inventions). Even though the clearly defined format of patent documents makes analysis easier, patent mining is still difficult. Here is when patent mining tasks and methods come into play. While tasks are precise actions that must be taken to complete the patent mining process, methods are ways that a job can be carried out.

To get more information, read entire article on Patent Mining