Using domain-content specific natural language processing models, we extract granular signals from text to generate detailed, data-driven insights. Besides your brand’s sentiment score, we can point out exactly what attributes are favoured or disliked.
Our proprietary 15 layers of data filtering framework cleans out spam and irrelevant data before the analysis stage. This ensures a high level of data veracity for our research.
We combine human expert knowledge with artificial intelligence to bring you actionable and forerunning insights. Our research, data science and insights teams develop domain specific ontology to enhance insight mining and trend detection capabilities.
We recognise new product attributes and measure their potential impact to the market.
Our deep learning models deliver both high level of predictive accuracy and explainability.Conventional deep learning models offer predictions, but they cannot explain why or how a phenomenon happens.
By integrating clean data, human expertise knowledge and our advanced explainable neural network technologies, we can identify the relationship between attributes, features, and people to understand the commercial impact and predict a future occurrence.
Besides understanding the context and semantics of text content, it is critical that the machine does it with high accuracy.6Estates has researched and developed various technologies to apply in the understanding of content in different formats such as financial documents, news reporting, online conversations etc.
We have developed our enhanced version of the latest and powerful pre-trained models BERT (Bidirectional Encoder Representations from Transformers) and ELMo (Embeddings from Language Models).
Sentiment Understanding & Product Attribute Mining Domain-assisted Hierarchical Organization of Product Consumer Reviews (11397N) and Product aspect ranking and its applications (11397N)
Unstructured to Structured Data Automatic FAQ compilation for community-based question-answering archive (09266N)
A Method and Apparatus for Tracking Microblog Messages for Relevancy to an Entity Identifiable by an Associated Text and an Image (13224N)
Sentiment Understanding & Product Attribute Mining Domain-assisted Hierarchical Organization of Product Consumer Reviews (11397N) and Product aspect ranking and its applications (11397N)
Singapore
Blk 79, Ayer Rajah Crescent, #01-07/08, Singapore 139955