What it is, how it works and how it can help you
What is TechDiscovery?
If you want to quickly explore a patent landscape for a specific technology space or look for patents that are similar to your examples, TechDiscovery is the right tool for you. It allows you to quickly get to the list of the most relevant families for your analysis and easily refine these if necessary. Skip the search part, and get straight to the data and analysis you require.
You can use either patent / family numbers or text to begin a search. The text can be either a simple description of the technology, or longer form technical information (e.g. your own unpublished patents, or even product descriptions from competitors. You can also use a combination of both patent examples and text (where possible).
The purpose of this type of search is to find a set of relevant patents to the technology or invention. These patents can then be useful for patent analysis to quickly and easily evaluate the largest players and trends within a tech space for example. Please note that TechDiscovery does not replace the accuracy and exhaustive nature of classifiers that's required when building landscape reports for example.
How to use TechDiscovery
Video tutorial
Step by step image walk through
1) From the Classification landing page, select 'TechDiscovery'.
You have the option when you first start using the TechDiscovery platform, to 'Don't save search history' or 'Save search history'. 'Don't save search history' means that you cannot access your own previous searches in your Classification account, whereas 'Save search history' means that you can.
If you select the option to save search history, the 'Pinned and Recent Searches' option operates like the report history and allows you to review your recent and saved TechDiscovery searches in the platform. You can also keep ad-hoc searches private by selecting 'Do not save searches' here. These will not show in your recent searches:
2) Next, enter your search criteria whether that's: • Patent numbers or family ID's • A Technology, for example, noise cancelling headphones • Paragraphs of Text, for example, a quote from a patent abstract, or a detailed description of an invention. Use a new row for each criteria type. Your search criteria is the basis for finding similar patents. In this example, I'm going to enter in two patent numbers and 'Noise cancelling headphones' as my technology.
3) Select 'Search'. By default, TechDiscovery will generate a list of 1000 similar patent families to the search criteria. Results are given a similarity score between 0 and 100, with the highest scoring patent families showing at the top.
4) You do have the ability to 'add more text' to your search retrospectively, should you wish to add a paragraph of text to the search for example (or add a technology name should your search have originally started with a paragraph of text).
This helps to refine the search. Here I'm going to add some text from an existing patent family which I deem to be relevant to the technology I'm looking for. Click 'Apply and update'.
5) Next, select patent families in the 'similar patent families' list that you decide are most relevant to the search criteria you entered, by clicking the 'plus' icon. This will help train the search with yet more good examples. Use the minus icon to simply remove non relevant examples from your results, or if you're unsure on certain examples, you can use the Review Later button to store these away.
You can do this based on the information you can see in the list view, or select the icon below to view the full patent family details. This will open in a separate tab for you to review and then return to the text explorer view.
Handy Hint: The 'magnifying glass' icon shown next to this allows you to start a new TechDiscovery search from that patent family.
The Boolean search bar at the top of the results (Shown above) can be used to filter results to specific keywords (see boolean search guide at the 'i' next to the search field) in order to, for example, add patent families to the relevant or irrelevant selections.
6) Select 'apply and update' to refine your results based on these updates. You can perform a couple of iterations of this process to generate more relevant results.
7) You then have the option to 'Create a report' from this list. This allows you to gain insights from the similar family data and review for instance, top owners in the technology space, or to review filing trends in the data for example.
8) You also have the option to 'export' the results of similar patent families. When selecting to 'export' you in fact, have these options:
- Download to CSV or Excel
Allows you to download the results whether you want to review various pieces of data such as first filing country or publication date; to share the list with others; or import to another platform.
- Start report from this list
Allows you to take the similar patent family list into a custom report to combine with other data, or to select your technology grouping options. Reports generated from the TechDiscovery can be found in your report history.
- Cluster this list
This option automatically triggers a new report to be generated from the similar patent family list. Results will be clustered.
- Build UTT report from this list
This option automatically triggers a new report to be generated from the similar patent family list and results will be grouped using the Universal Technology Taxonomy. You require the UTT in your account to have this option available to you.
- Copy Family IDs to clipboard
Provides the option to copy the patent family ID's to your clipboard, in case you want to paste to a patent upload report for example.
9) Remember, you had the option when you first start using the Tech Explore platform, to save the searches or keep them private. If you chose to save your searches, the 'Pinned and Recent Searches' option allows you to review your recent and saved TechDiscovery searches in the platform. All searches are saved for 30 days but if you 'pin' your search it will be saved in your pinned searches.
10) If you wish to change these preferences, they can be amended in your settings via the cog icon here:
How does TechDiscovery work?
Classification uses a very sophisticated proprietary patent linguistic algorithm that has been tried and tested over the past two years across our Universal Technology Taxonomy (“UTT”). It is more advanced than most other systems on the market and will typically therefore provide better results than other similarity search tools available on the market.
Similarity searching starts with vectorising every patent family in the universe (think of this like giving each patent family a unique fingerprint). Each patent family can then have its vector (fingerprint) compared against others to identify vectors (patent families) that is closest to it, returning the closest results based on the chosen sample size (50, 100, 1000 etc.). Classification’s deep learning model (“algorithm”) is specifically designed for patent linguistic tasks and uses the patent title, abstract and claims to generate a vector for each individual patent family. Classification uses generative AI to assist the vectorisation (or fingerprinting) of information when you have very little information e.g. a technology name or brief description.
How does TechDiscovery differ from Semantic Search?
Here are the key differences to semantic searching/boolean searching in the Classification platform:
- The results are much better (more relevant results, more results found)
- You can search based on patent numbers
- You can use whole paragraphs of text
- The search is specific to patent text, not 'general text similarity'
- You can combine multiple inputs into one search for example, patents, blocks of text, and technology names
How can TechDiscovery help you?
This feature could assist you when you are trying to understand a technology area quickly or where you have very limited information. You might want to understand more about a specific product and what is out there that is similar, or it may assist you in prior art searches. TechDiscovery is very explorative, and only needs a small amount of time to generate results which can be beneficial. Classifiers on the other hand provide the most accurate results, are repeatable and are exhaustive, which is ideal for generating landscape reports.
TechDiscovery could be used to explore a technology area before deciding whether to build a classifier for that specific area. Do note however that Classifiers will be more accurate and once built, they sit in your account and can be used repeatedly in your reporting and across teams. With any questions about TechDiscovery, please contact a member of the Classification Team.