Creating a New Entity Recognizer 

In Business, you often need to extract key entities from information obtained via various sources which necessarily need not be in a pre-defined structure. The Smart Bot provides a simplified solution to the complex task of preparing a dataset for entity recognition and training a custom recognizer (model) that uses AI and ML.

To create a new Entity Recognizer, you need dataset. So there are two steps in creating a new Entity Recognizer:

• Preparing Datasets
• Creating Entity Recognizer Model

Preparing Datasets

1. Navigate to Smart Bot > Entity Recognition.

The Entity Recognition page is displayed with two tabs - namely Dataset and Model.

2. Open the Dataset page.
3. Click Add Dataset. 

4. Enter a unique Dataset Name. It should start with a letter and can contain only letters, numbers, space and underscore.
5. Optionally, enter an additional Description.
6. Select/browse Dataset file from Upload File option. The selected file must be in the ".CSV" (Comma Separated Value) or ".xlsx" ( Microsoft Excel Open XML Spreadsheet), where the first column is a narrative text and the other columns represent the entities to be extracted.

The guidelines to prepare a training dataset are the same as presented in the Text Classification. The following image shows a sample training dataset to extract Service Provider and Client form deal or contract notes.

When your entity values are repeated in the text, then you can provide the exact location of your entities in your dataset. You can prepare such dataset with different NER annotation tools in the form of json format and upload it for training. 

Model page helps to select uploaded datasets from which you can train the model.

Creating  Entity Recognizer Models

1. Navigate to Smart Bot > Entity recognition.

The Entity Recognition page is displayed with two tabs - namely Dataset and Model.

2. Open the Model page.
3. Click Add Model.  

The Classifier Configuration window is displayed.

4. Select Create.
5. Enter the unique Entity Recognizer Name. It should start with a letter and can contain only letters, numbers, and underscore.
6. Select the required dataset from the Dataset drop-down.
7. Select Method option from the drop-down. The supported methods are:
   • Bert
   •  Spacy

This selection is optional and are for advanced user. The default method is Bert.

8. Optionally enter additional Description.
9. Click Submit.

A model is created in the list with the status ‘In Progress’. The Smart Bot will take some time to train the recognizer based on the size of the dataset. When the training is completed, Smart Bot will update the status to 'Completed'.