One stop solution of text mining, public opinion, dialogue and other functions with the industry-leading natural language analysis technology; Based on various of natural language core algorithms, we fully meet the needs of various natural language understanding and help the product to understand the world.
• According to the grammar rules of the source language, the corresponding grammatical categories are identified from the results of the lexical analysis, and the grammar check is performed at the same time.
• According to different scenarios, to apply functions such as analysis, word property tagging.
• Analyze the core information of an article, extract keywords, automatically classify the articles according to the content, and provide support for article content analysis;
• The article tag service provides an in-depth analysis of the article title and content, and outputs multi-dimensional tags, such as themes, topics, entities, etc., and the corresponding confidence levels. These tags reflect the key information of an article. This technology has high practical value in scenarios such as personalized recommendation, article aggregation, content retrieval.
• Based on domain ontology information, the semantic similarity calculation of text data is studied from three aspects: concept, statement and document;
• Described the semantic extraction, semantic description and semantic computing in detail.
• Including concept similarity/correlation calculation method, statement similarity calculation method, document similarity calculation method, indexing and sorting technology of domain search engine based on semantic.
Word Vector Representation
• Based on KANKAN AI owned word vector and deep learning, it can complete the word vector representation quickly, in which the words or phrases from the vocabulary are mapped to the real number vector. The methods for generating such mapping include neural network, dimensional reduction of word mutualistic matrix, probability model, interpretable knowledge base methods, and explicit representations of terms and background of the word.
• When used as a low-level input representation, word and phrase embedding have been proven to be able to improve the performance of NLP tasks, such as grammar analysis and emotion analysis.
• It is the process of analyzing, processing, summarizing and reasoning subjective texts with emotional.
• Emotional analysis or opinion mining is people's perspectives, emotions, and attitudes toward entities such as products, services, organizations, and so on. The development and fast start of the field benefited from social media on the web, such as product reviews, forum discussions, Weibo, and the rapid development of WeChat, because this is the first time in human history to have such a huge digital form record.