Here are four exemplary usage scenario NLP interfaces and services.
(1) Text intelligent error correction
Smart customer service jobs in the scene , for voice information input by the user in the process of speech-to-text (ASR → TTS) in, inevitably a lot of dislocation context and text of the error appears on the wording. Faced with these errors, if you want a manual to be treated, it consumes a lot of labor costs. In this case, the use of syntactic analysis and dependency interfaces text correction , semantic association between each language text units for analysis, while the automatic correction of the text. The function of the text by intelligent error correction, can efficiently assist labor, effectively improve the quality of speech-to-text characters.
In addition to customer service in the smart user input checking the scene, in the image-to-text publications or scanned copy of the document (OCR) scenes , voice-to-text error conditions similar intelligent customer service is also appear. Thus, syntactic analysis and dependency text correction interfaces can also be used to verify the character recognition result in the image's.
In summary, the smart customer service to check user input, image character recognition result verification, audit office documents, quality control and other intelligent text demand scenario, the text has intelligent error correction. Therefore, text correction and syntactic dependency analysis interface is widely used in these scenarios.
(2) extracting structured financial documents
In the financial sector and government banking, insurance, securities, etc., there is often a large number of documents and documents need to manually sort, refine and archiving, the work takes a lot of manpower and time. In this case, you can use lexical analysis (intelligent segmentation, POS tagging, named entity recognition), keyword extraction and text categorization interfaces , first on a large number of financial or government document classification, and then the text data and structured information extraction key arrangement, can effectively assist human to handle various documents and documents, reduce labor costs.
(3) network text retrieval
For text data in the database or the Internet environment, different users have different information needs, they will be searched or retrieved from the corresponding text data according to their specific needs. In this case, you can use lexical analysis (intelligent segmentation, POS tagging, named entity recognition) and keyword extraction interfaces , title and body text are the core keyword analysis to extract key information to reflect the text of the theme, topic, entity, etc. multidimensional vocabulary label; and then polymerizing the same keyword text label data, and provides precise user-specific content to be a search, the user-friendly type of text information to read a full range of data.
For example: content search, Wanfang database, Chinese HowNet information flow and other literature search
(4) Article key information extraction and content analysis
Use lexical analysis (intelligent segmentation, POS tagging, named entity recognition), text classification and keyword extraction interfaces , the first article Subject classification by content type (such as the article belongs to a topic technology, entertainment, sports, finance and so on, AI , film, football, stock and other two topics), and then to the title and content of the article are the core keyword extraction, output may reflect the multi-dimensional articles critical information label and vocabulary, finally reached a vertical type of information resource development, text content analysis the purpose of all types of content and applications to best meet the needs of the user.
[Link] Quick Experience
Tencent cloud natural language processing API documentation: https://cloud.tencent.com/document/product/271/35484