alt Inc. (Headquartered in Tokyo, Japan and represented by Mr. Kazutaka Yonekura as CEO) which has developed al+ P.A.I. (Personal Artificial Intelligence) has won the first prize at the 2018 Named Entity Recognition Contest held by the Vietnamese Language and Speech Processing Conference (hereafter referred as “VLSP”) on March 23, 2018 at University of Science and Technology of Hanoi in Vietnam.
VLSP is the largest natural language processing and speech processing conference which gathers Vietnamese researchers in this field. In the conference, there are various contests in the domain of natural language processing and speech processing including Vietnamese named entity recognition, sentiment analysis, speech synthesis and speech recognition contests. In the named entity recognition contest, each team has to train their machine learning model using the training data and report the result predicted by their system on the test data provided by organizers of the conference. Participation teams are ranked by the accuracy scores of their models, which are calculated on the test data set.
This year, alt team led by Dr. Pham Quang Nhat Minh from alt AI R&D division, participated in the Vietnamese named entity contest and the sentiment analysis contest. alt won the first prize in the named entity recognition contest and third prize in the sentiment analysis contest.
Dr. Minh implemented a named-entity recognition model that can recognize nested named entities and received high praises from the conference committee. Nested name entity recognition is a more difficult task than general name entity recognition, and it is challenging even to human beings in some cases.
Following the victory of alt’s joint researcher, Dr. Le Hong Phuong at the previous VLSP 2016 in the same contest, this will be the second consecutive win of alt.
At alt, to accomplish highly accurate P.A.I. system, we have based on most significant research work on natural language processing, and we will continue to conduct world-level studies in AI and NLP fields.