Natural Language Processing

With fully independent intellectual property rights, Xiao-i possesses multilingual natural language processing capabilities, covering words, phrases, sentences, text and other aspects of the extraction, analysis, text mining and reasoning ability. It meets the needs of enterprises and professional users in human-machine interaction, text mining, information extraction and knowledge construction.
Features: Independent intellectual property, multi-lingual, precise understanding
Scenes: Knowledge graph, information searching, document comparison

Deep Semantic Interaction

With flexible and diverse session management and context processing techniques that take full advantage of deep and enhanced learning techniques, deep-level scene session processing, intent recognition and complex logic reasoning are achieved. Structured knowledge and analytical language understanding are combined as well. The business value of conversational AI is thus realized in a wider range of applications.
Features: Contextual understanding, multi-turn dialogue, logic reasoning
Scenes: Customer service bot, task-oriented conversations, product recommendation

Machine Learning

Research and master all kinds of traditional machine learning and the latest approaches, including deep learning, enhanced learning, transfer learning, GAN, etc. Those machine learning approaches are applied to fields such as natural language processing, speech recognition and image recognition. In business scenes including precise marketing, personalized recommendation and risk control, big data and distributed processing algorithm are combined, forming an efficient human-machine collaboration learning system.
Features: Neural network, enhanced learning, big data
Scenes: Precise marketing, risk management, stock prediction

Affective Computing

Deep learning technologies are deployed to identify, understand, process, and simulate human emotions. The affective computing capabilities in dimensions including text, speech, facial expression and human body posture are realized. Also, the capabilities of emotion interaction processing and real-time dynamic affective computing and analysis have been developed, widely applied in various practical business scenarios.
Features: Simulated emotion, self-learning, sentiment interaction
Scenes: Welcoming bot, service bot, companion robot

Speech and Image

Fully self-developed speech recognition, speech synthesis and voiceprint recognition technologies have been widely used in a variety of business scenarios. The unique computer vision system already has OCR capabilities for bank cards, business licenses, document forms, ID cards, etc. At the same time, it has the ability of scene analysis such as face recognition and analysis, multi-target tracking, pose recognition, semantic and entity segmentation. Through the self-developed deep learning framework for rapid engineering, with ultra-high concurrency, can be quickly targeted to the industry areas for optimization.
Features: Business application, high concurrency, targeted tuning
Scenes: Outbound call bot, OCR recognition, smart security