The Cross-enterprise Lead Mining application helps you mine user profiles and identify upsell opportunities from sources like social media data and call logs. It also gathers sales leads related to competitors and industry trends.
Access
Log on to Alibaba Cloud Model Studio (Bailian). On the Application Gallery page, click Cross-enterprise Lead Mining to enter the console of this lightweight application.
Features
Application details
In the Cross-enterprise Lead Mining application, on the Application Details tab, you can view five sections: Function Description, Target Audience, Best Practices, Billing Rules, and Quanmiao-related Application Recommendations.
These sections provide more details: The feature description explains how to mine user profiles and upsell opportunities from social media data and call logs. The target audience is the marketing and sales teams of enterprise customers. The best practices demonstrate how to generate tagged results by applying multi-level tags to materials such as chat or call logs. The billing models include Qwen-Plus-Latest (¥0.0008 per 1,000 input tokens, ¥0.002 per 1,000 output tokens), Qwen-Max-Latest (¥0.0024 per 1,000 input tokens, ¥0.0096 per 1,000 output tokens), and Qwen-Long (¥0.0005 per 1,000 input tokens, ¥0.002 per 1,000 output tokens). Recommended related applications include Cross-enterprise VOC Mining, Web Content Moderation, and Miaocè.
Result tuning
Click the Effect Debugging tab to set the configuration items by following the steps below. On the application page, click the Effect Debugging tab.
Step 1: Upload material
-
Paste your source content, such as a conversation or a call log, into the text box. Alternatively, to use a pre-filled example, go to the Sales Leads tab, select a scenario, and click Use this example to pre-fill the task description.
-
Select a model for the clues. The default is Qwen-Max. In the Select Model drop-down list, select the model that you want to use. The options include Qwen-Max, Qwen-Plus, Qwen-Max-Latest, Qwen-Plus-Latest, and Qwen-Long. After the configuration is complete, click Start Mining.
Step 2: Configure content tags
-
Enter a Task Description. In the Task Description text box, enter a description for the annotation task. The description can contain a maximum of 500 characters.
-
Customize lead tags.
-
Enter the tag name, meaning, and values in the specified format. You can input tags individually or in batches. Refer to the official Example for guidance. Click Example to open the example panel on the right. You can select a preset scenario (such as real estate sales analysis, job lead analysis, or user profile analysis), select the desired tags, and then click Append Tags to add them to the tag definition list.
-
Click Batch Input to paste all tag names, meanings, and values at once for automatic recognition. This form also includes a task description input box (up to 500 characters) and an Add business knowledge switch. On the right-side panel, you can paste tag information for automatic recognition. Each line represents one tag, with the tag name and its values separated by a delimiter. You can choose a Tab (for Excel), @@@, or a Custom delimiter. After pasting, click Recognize and view in the left panel to complete the parsing.
-
Click Add Custom Tag to define custom tags for more precise mining.
In the New Tag dialog box that appears, enter a Tag Name (up to 20 characters) and the Tag Meaning and Values (up to 2,000 characters), then click OK.
-
-
Optional: Add business knowledge.
In this text box, enter supplementary information, such as business context, task-related notes, or examples.
In the Add business knowledge section, enable the switch on the right (it turns purple), and then enter content in the text box below.
Step 3: Set the output format
You can define the output format based on the Format Example on the right, or click Auto-generate JSON Format within the input box to generate the format automatically.
The Format Example panel contains json and text tabs with output templates for each format. After confirming an example, click Replace with this example to populate the output format input box with the example's content.
Step 4: Start mining
Click the Start Mining button. After the mining process is complete, the structured JSON results are displayed on the right. The JSON output contains the configured fields and their extracted values, for example: IsRecentHomeBuyer=Y, PaymentMethod=CommercialLoanOnly, FamilySize=AtLeast3, RoomCount=3-bedroom, PurchasePurpose=Self-use, CustomerType=Upgrade, Budget=Approx1.5M, and CoreNeeds=[SchoolDistrict]. The top of the right panel provides a Show JSON Block drop-down menu, a Download as Excel button, and a Copy button. The bottom of the panel displays usage information, such as "input tokens: 696" and "output tokens: 110".
API examples
After you finish debugging the effect, click the API tab to view the corresponding generated API examples.
The API supports two invocation methods:
-
Real-time streaming API: Processes a single request (one text input per call) and returns the result in real time as a stream. The streaming response protocol is Server-Sent Events (SSE).
-
Batch processing API: Processes a submitted file or multiple text segments. A background task is created, and you must poll the task status by using the task ID. The results are returned after the task is complete.
The API tab contains code examples for Java and Python. The Java example uses the org.example package and requires dependencies such as fastjson2, Alibaba Cloud auth credentials, the SDK gateway, and quanmiaolightapp20240801.AsyncClient.