Incremental training

更新时间:
复制 MD 格式

Incremental training for Product Review Parsing

Incremental training lets you add custom labels to the platform's pre-trained models for product review parsing. These models cover the E-commerce, automotive, and local life domains. You can then train a complete parsing model using only the new labels.

If the supported domains and industries do not meet your requirements, you can join the NLP Self-Learning Platform Q&A Group 2 (DingTalk group ID: 44619071) for assistance.

Activate the service and purchase a resource plan

You must activate the service before you can train models on the NLP Self-Learning Platform. After you activate the service, you can purchase discounted resource plans.

NLP Self-Learning Platform: Activation page

NLP Self-Learning Platform resource plan: Purchase page

Procedure

1. Model creation

In the Create Project section, select Product Review Parsing to create a project.

Enter the project information and click OK.

2. Data annotation

In My Projects, go to the Data Center to manage your data. You can create data in two ways:

1. Create an annotation task. 2. Upload a dataset.

2.1 Create an annotation task

Step 1: Upload documents for annotation and add annotators

Note

The project creator and project administrator are annotators by default. You can also assign annotation tasks to your RAM users. These users can then log on to the platform with their RAM user credentials to perform data annotation.

RAM user logon instructions:

1. Go to the RAM user logon page: https://signin.aliyun.com/login.htm

2. After you log on, go to My Projects to perform the annotation task.

Note: Only data files that use UTF-8 encoding are supported.

Step 2: Set the review categories to addStep 3: Annotate data

2.2 Upload a dataset

Alternatively, you can upload training data that has been annotated locally. You must format the data according to the sample file before you upload it.

3. Create a model

In Model Hub, click Create Model.

Enter model information

Set Model Type to Product Review Parsing - Classification - High-Precision - Bert.

Available modes

  • Disabled: Only annotated data is used for training.

  • Fuse with platform data: Trains the model using your custom annotated data combined with pre-annotated data from 18 industries on the Alibaba E-commerce platform.

  • Incremental only: Trains the model using your custom annotated data for the E-commerce, local life, and automotive domains.