SpaCy NER Model Training Guide

Step 1: Upload Your Resume File

Upload a resume or document file for text extraction. Supported formats include:

• PDF
• DOCX (Word Document)
• RSF (Rich Structured Format)
• ODT (Open Document Text)
• PNG, JPG, JPEG (Image Formats)
• JSON

Ensure that your file is in one of the supported formats before uploading. The system will extract and process the text from your document automatically.

Proceed to Upload

Step 2: Preview and Edit Extracted Text

After uploading your document, you will be shown a preview of the extracted text. This preview allows you to edit the text if needed to correct any extraction errors or remove unwanted content. Once you're satisfied, click "Next" to proceed to Named Entity Recognition (NER) annotations.

Proceed to Text Preview

Step 3: Annotate Named Entities

In this step, you will preview the Named Entity Recognition (NER) results generated from your text. You can add new entity labels, select relevant text for each label, and make manual adjustments. Once you’ve annotated the text with the appropriate labels, save your annotations and export the data in JSON format for model training. NOTE:(following labels can be taken in use: ["ABOUT","CERTIFICATE", "COMPANY","CONTACT","COURSE", "DOB", "EMAIL", "EXPERIENCE", "HOBBIES", "INSTITUTE", "JOB_TITLE", "LANGUAGE", "LAST_QUALIFICATION_YEAR", "LINK", "LOCATION", "PERSON", "PROJECTS", "QUALIFICATION", "SCHOOL", "SKILL", "SOFT_SKILL", "UNIVERSITY", "YEARS_EXPERIENCE"])

Instructions:

• Click "Begin!" to load the extracted text.
• Highlight sections of the text and assign them to the available labels.
• Add new labels if necessary.
• Once done, click "Export" to download your annotations as a JSON file.

Proceed to NER Annotation

Step 4: Save and Format JSON Data

Upload your annotated JSON file from the previous step. The system will process and reformat the JSON file to ensure compatibility with the SpaCy model training process. After formatting, you can proceed to the model training step.

Instructions:

• Upload the JSON file you downloaded after the annotation step.
• Click "Process" to reformat the file.
• Once processing is complete, click "Next" to proceed with training.

Proceed to Save JSON

Step 5: Train the NER Model

In this final step, you will convert the formatted JSON data into the SpaCy format and begin training the NER model. You can customize the training by selecting the number of epochs (iterations) the model will go through and setting the version for the trained model.

Guidelines:

• Number of epochs: The higher the number of epochs, the more times the model will learn from the data, but too many epochs can lead to overfitting. Start with 10 epochs for a balanced training approach.
• Model versioning: Provide a version name for this training session, so you can keep track of different versions of the model.

Once the training is complete, you can download the latest version of the trained model for use in production.

Proceed to Model Training