Cognitive Analytics Overview
Introduction of Cognitive Analytics
There are two major meanings of the word: cognitive, which stands for the act of knowing or the mental processes of perception. While the Cognitive Network Toolkit targets on creating intelligent machines for the act of knowing, this Cognitive Analytics Toolkit targets on using machines to detect, perceive, or predict human's perception.
IBM System G Cognitive Analytics Toolkit provides a wide range of tools to detect human's emotion and perception on text, images or videos.
Based on Visual Analysis and Machine Learning techniques, from hundreds of millions of social multimedia content of tagged images, we created thousands of visual object detectors and thousands of visual adjective detectors. By combining these adjective-noun pairs, it became possible to detect the visual sentiments in an image or video. For instance, we want to detect "beautiful flower", "crying baby", "crazy car", "lonely person", etc. These tools help designers to predict the end users' responses and by leveraging the prediction results, the marketing/information spreading stragtegies can be more effective.
Based on Natural Language Processing and Machine Learning techniques, we created Text Emotion Detectors which allows machines to detect disgruntlement, depression, anxiety, etc, to help understand a person is under stress. We identify not only positive, neutral, and negative sentiments but also fine-grained emotions from user-generated content. Combining with the Behavioral Analytics, this may be used for reducing the risk of mental breakdown.
(1) Visual Sentiment and Recognition
A set of tools that detects visual objects, adjective-nouns pairs appeared in an image or video, and the predicting visual feeling aroused by the users. It can be used for detecting the feeling the photographer or director wants to convey. It can be used for detecting emotions the viewers may feel. It may even provide automatic comment (on image or video) suggestion for the viewers.
(2) Text Emotion Analysis
A fine-grained emotion detection from text. This tool was based on supervised learning on annotated user generated content in public domain social media domain and Enron email dataset.