Thomson Reuters is a global company that supplies dependable data and information to a variety of businesses. The firm primarily services the legal, tax, and accounting professions, as well as the media. Artificial intelligence is becoming the standard, and businesses are quickly adopting it.
Businesses may achieve efficiency and development by utilizing AI and other disruptive technologies such as machine learning. Companies are heavily investing in AI and developing new products. LG AI Research has spent $89 million on a project dubbed “Super Giant AI.” Artificial intelligence is also a priority for Thomson Reuters.
Trust, security, privacy, and a human-centric approach are among the AI principles in place at the organization. Some of their study topics include natural language, machine learning, and information retrieval. Some of their AI research initiatives are listed below.
Examining documents in a more coherent way
Thomson Reuters is working on an artificial intelligence initiative aimed at making document analysis and development easier for knowledge workers. They’re looking for segmentation algorithms that can extract structural cues from a page and break it into headings, subheadings, and so on for this to work.
Then there’s NLP-based entity and relation extraction, which can figure out how various pieces of a document are related. This is especially important when it comes to financial and legal paperwork. Developing a taxonomy and constructing models might take time, so defining a study topic may be more convenient.
Active learning, hybrid machine learning models, and other techniques are included. Through this study effort, the corporation hopes to analyze deviations and synthesize the product to meet the needs of each consumer.
Text mining powered by machine learning
Thomson Reuters used machine learning and natural language processing (NLP) models to tackle this problem since text mining and accurate information extraction are critical for tax and legal systems. This increases the value of analytical products. The two most important phases in this research are motion analysis and party result identification.
Thomson Reuters can provide more clients with valuable information thanks to its machine-learning-based methodology.
Readability of Machines
Thomson is heavily investing in this AI initiative, which will use deep learning to experiment with and create futuristic methods to machine reading comprehension challenges. These models can subsequently be used in the company’s relevant initiatives as well as in question-answering situations.
The firm is experimenting with BERT and claims that it has delivered outstanding outcomes for legal and tax-related duties thus far. The company’s long-term goal is to create a conversational system that can interact with consumers in real-time and offer them the information they require.