In this article, we wanted to give a brief overview of how you can add an A.I. model that works with text. There are many things you can do with text analysis - detect sentiment, analyze key phrases for a summary, find out more context with entity detection, and etc.
To create a text analysis model go to AI models . The list of available models will appear.
2. Press the Add button and the creation model screen will appear. Start by filling in the general information.
3. Next, choose what do you want to analyze - Text or Images.
4. Then, we need to select what we want to detect.
5. Then we need to specify the Model Operation mode - Pass/Not Pass or Detection Mode. The first mode is a kind of Yes or No answer if the information is in the text; the detection mode just detects labels.
6. Then you need to specify the Model service - Entity or Sentiments or Key Phrases.
7. In our example we will use Entities. The entity is a term that is recognized by an algorithm like an organization name in the context (apple vs Apple Inc) or a person name. For instance, you can use entity=date to detect dates in application forms or checklists; also you can detect the insurance claim number (you will need a custom algorithm for that) as an entity from the document. Once you select the service, the extra screen will appear.
8. Next, select the behavior: Pass/Not pass or Detection mode.
In the example above, if you select Pass/Not pass mode, then any incoming document that has any of the above entities will be considered approved.
9. Now, choose how you want to treat the content.
The difference here is that if you have two documents to analyze at once, then Per item means that the entities will be checked per each document separately. And if one of them does not have any entity, then both are rejected.
10. Once you are done press Apply. The model will appear in the list.