Product introduction
Introduction
On December 31, 2020, the basic NLP service was commercially launched. In March 2023, the basic NLP service was fully upgraded to Natural Language Processing (NLP).
Compared to versions 1.0 and 2.0 of the basic NLP service, NLP has the following updates:
NLP is now organized into five main categories: Basic Text Services, E-commerce Industry Capabilities, Conversational Intelligence Services, General Industry Capabilities, and Healthcare Industry Capabilities.
The pre-trained models from the original NLP Self-Learning Platform have been moved to E-commerce Industry Capabilities and General Industry Capabilities. This helps users quickly call the corresponding industry services through APIs.
A new console interface provides a better experience for managing created services. Access the console.
The pricing plan has been upgraded to provide each user with a larger free quota for calls.
Product calls: To make calls, see Get started.
1. Basic text services
1.1 Basic Edition
Algorithm service | Supported realm | API information | Free quota | Default QPS | Maximum text length |
Chinese tokenization | General | 500,000 calls/day | 20 | 1024 characters | |
Part-of-speech tagging | General | 500,000 calls/day | 20 | 1024 characters | |
Named entity recognition | E-commerce | 500,000 calls/day | 20 | 1024 characters | |
Sentiment analysis | General | 500,000 calls/day | 20 | 1000 characters | |
Keyword extraction | Chinese e-commerce | 500,000 calls/day | 20 | 128 characters | |
English e-commerce | 500,000 calls/day | 20 | 40 characters | ||
Smart text classification | News | 500,000 calls/day | 20 | 5000 characters | |
E-commerce | 500,000 calls/day | 20 | 5000 characters | ||
Word embedding | General | 500,000 calls/day | 20 | 1024 characters |
1.2 Premium Edition
Algorithm service | Supported Realms | API information | Free quota | Default QPS | Maximum text length |
Multi-language tokenization | General | 500,000 calls in total | 20 | 1024 characters | |
E-commerce title | Multi-language Tokenization Premium Edition (E-commerce Title) | A total of 500,000 times | 20 | 1024 characters | |
E-commerce content | Multi-language Tokenization Premium Edition (E-commerce Content) | 500,000 times in total | 20 | 1024 characters | |
E-commerce review | Multi-language Tokenization Premium Edition (E-commerce Review) | A running total of 500,000 times | 20 | 1024 characters | |
Entertainment | Total: 500,000 times | 20 | 1024 characters | ||
O2O | A cumulative total of 500,000 times | 20 | 1024 characters | ||
General Southeast Asian languages | Multi-language Tokenization Premium Edition (General Southeast Asian Languages) | A total of 500,000 times | 20 | 1024 characters | |
E-commerce Southeast Asian languages | Multi-language Tokenization Premium Edition (E-commerce Southeast Asian Languages) | A total of 500,000 times | 20 | 1024 characters | |
Named entity recognition | E-commerce | Total: 500,000 times | 20 | Chinese: 1024 characters Multi-language: 500 characters | |
E-commerce Southeast Asian languages | Named Entity Recognition Premium Edition (E-commerce Southeast Asian Languages) | Total: 500,000 times | 20 | Chinese: 1024 characters Multi-language: 500 characters | |
Part-of-speech tagging | E-commerce | 500,000 times in total | 20 | 1024 characters | |
Word embedding | E-commerce | 500,000 calls in total | 20 | 1024 characters | |
Review | A cumulative total of 500,000 | 20 | 1024 characters | ||
Search | 500,000 times in total | 20 | 1024 characters | ||
Entertainment | A cumulative total of 500,000 times | 20 | 1024 characters | ||
Sentiment analysis | E-commerce Southeast Asian languages | Sentiment Analysis Premium Edition (E-commerce Southeast Asian Languages) | 500,000 calls in total | 20 | 500 characters |
Text correction | General English | Total: 500,000 times | 20 | English: 512 characters | |
General Chinese | Total: 500,000 times | 20 | Chinese: 128 characters English: 500 characters | ||
Text summarization | E-commerce | Total: 500,000 times | 20 | Minimum length of 60 characters and maximum length of 800 characters | |
Text similarity | E-commerce | Total: 500,000 times | 20 | 512 characters | |
Dependency parsing | E-commerce | 500,000 calls in total | 20 | 120 characters | |
General (CTB tags) | Up to 500,000 times | 20 | 120 characters | ||
General (Stanford tags) | 500,000 calls in total | 20 | 120 characters | ||
Table Q&A | General | Free during public preview | 10 | 120 characters |
2. E-commerce industry capabilities
Algorithm service | Supported realm | API information | Free quota | Default QPS | Maximum text length |
Smart product publishing | Category prediction | 500 times in total | 20 | 800 characters | |
Price extraction | 500 times in total | 20 | 800 characters | ||
Brand prediction | Up to 500 times | 20 | 800 characters | ||
Product review parsing | E-commerce realm | A total of 500 calls shared with General Industry Capabilities | 10 | 500 characters | |
Local services | 10 | 500 characters | |||
Automotive realm | 10 | 500 characters | |||
Product review purchase decision analysis | E-commerce realm | 10 | 500 characters | ||
Automotive realm | 10 | 500 characters | |||
E-commerce customer service | Online customer service | 10 | / | ||
Online customer service | 10 | / | |||
Other e-commerce capabilities | E-commerce realm | 10 | 500 characters |
3. Conversational intelligence services
Algorithm service | Supported realm | API information | Free quota | Default QPS | Maximum text length |
Table Q&A | General | Free during public preview | 10 | / |
4. General industry capabilities
Model name | Model description | Free quota | Default QPS | Maximum text length |
Classifies bidding announcements. It supports two types: "tender" and "bid-winning". | Total 500 times | 10 | / | |
Extracts 13 fields from bidding information, such as project name, project number, bidder name, and winning bid amount. | 10 | / | ||
Parses tender and bid-winning information separately. Extracts 22 fields from tender information. | 10 | / | ||
Parses tender and bid-winning information separately. Extracts 29 fields from bid-winning information. | 10 | / | ||
Extracts common elements from contracts. It supports 26 general element fields. | 10 | / | ||
Extracts information that follows a key-value pattern from documents such as resumes, contracts, and reports. | 10 | / | ||
Suitable for outbound telesales calls. It classifies conversations by industry and scenario for applications such as voice quality inspection. It supports over 30 industries and over 170 scenarios. | 10 | / | ||
Supports scenarios such as customer service quality inspection in telesales and streamer monitoring in live streaming. | 10 | / | ||
Suitable for manual or smart outbound telesales calls. It recognizes the user's intent or reaction. | 10 | / | ||
Suitable for outbound telesales calls. It detects conversations with potential fraud risks for applications such as voice quality inspection. | 10 | / | ||
Extracts 10 fields from English resumes, such as name, contact information, degree, company, and position. | 10 | / | ||
Extracts 33 fields from Chinese resumes, such as name, gender, age, education, and employer. | 10 | / | ||
Extracts events from English news articles, covering 33 event categories. | 10 | / | ||
Predicts the category of a product based on its title in e-commerce scenarios. The category system is consistent with platforms like Taobao. | 10 | / | ||
Detects pornographic or erotic content in Chinese novels for content moderation. It outputs a confidence level for pornography and the related text. | 10 | 600 characters | ||
Analyzes short texts from social media in e-commerce scenarios to predict the sentiment of Russian text. Sentiments are classified as positive, neutral, or negative. | 10 | / | ||
Analyzes short texts from social media in e-commerce scenarios to predict the sentiment of English text. Sentiments are classified as positive, neutral, or negative. | 10 | / | ||
Analyzes short texts from social media in e-commerce scenarios to predict the sentiment of Spanish text. Sentiments are classified as positive, neutral, or negative. | 10 | / | ||
Supports scenarios such as telesales and online customer service. It recognizes the emotions of customers or agents, supporting 8 general emotions and 3 common business-related emotions. | 10 | 1000 characters | ||
Classifies one or more news texts. | 10 | / | ||
Suitable for live streaming scenarios. It uses ASR to convert speech to text and identifies poorly readable text caused by multiple people speaking at once. | 10 | 600 characters | ||
Parses documents for 10 causes of action and extracts 38 fields. | 10 | / | ||
Extracts keywords or summaries from documents. | 10 | 500 characters | ||
Designed for common text generation needs in real-world scenarios. It is suitable for generating text summaries or article titles. | 10 | 500 characters | ||
Generates in-vehicle startup welcome messages based on given weather information fields. | 10 | 500 characters | ||
Takes Chinese text as input and outputs its corresponding vector representation. | 10 | / | ||
Detects rhetorical devices like irony and sarcasm in social media. | 10 | / | ||
Analyzes humorous content to predict its type (such as pun, convention, or inversion). | 10 | / |
5. Healthcare industry capabilities
Algorithm service | Supported realm | API information | Free quota | Default QPS | Maximum text length |
Medical text analytics | Diagnosis normalization | Up to 500 times | 20 | 50 characters | |
Drug normalization | Up to 500 times | 20 | / | ||
Procedure normalization | 500 times in total | 20 | 50 characters | ||
Medical record deduplication | Total: 500 times | 20 | / | ||
Medical named entity recognition | 500 calls in total | 20 | 256 characters | ||
Medical text similarity | 500 times in total | 20 | 32 characters |