We have a feature called AI custom field tagging. This allows you to apply AI tagging for existing drop-down menu and drop-down tree type custom fields.
The algorithm considers the custom field values or nodes that have already been applied to patent documents, using this as training data for the AI algorithm, to tag selected patents and documents automatically.
The AI deep learning model (Fast Text and TextCNN) studies the training data, taking in consideration of the Title Abstract and Claims keyword, application domain and tech topic fields to understand the meaning of the custom field value or node.
With this calculation, the custom field value is applied to the selected records.
As this is an AI model, we recommend a higher volume of existing data (existing custom fields/tagging) to increase the tagging accuracy.
This AI custom field can be applied to a maximum of 50k patent at one time and is available for our Premium Account holders.
To use AI custom field tagging, you will have to start at the Workspace you wish to use it on.
Here you will be able to locate the 'Automatic Custom Field Tagging' button.
Then you will have the option of selecting a tagging method and which custom fields you would like to include in the tagging.
The training data consists of patent documents that have already been tagged with the custom field value/node. A minimum of 10 patents need to have the desired value/node applied to them in order to train the AI. The higher the volume of training data, the greater the accuracy of the AI auto-tagging algorithm.
Once you are happy, hit the "Start tagging" button at the bottom right of your screen, then click "Verify All Tagging Results" to view the results.
From here you'll then be able to verify the results of the AI tagging, by selecting the desired individual records and pressing "Confirm tagging" in the bottom-right pop-up menu.
After confirming the tagging, you'll then receive a confirmation that the patents have been verified.
