Third Sohu Content Recognition Algorithm Competition Concludes with Highlights and Insights

On May 24, 2019, the third Sohu Content Recognition Algorithm Competition concluded after 65 days, attracting 218 universities, 982 teams, and over 1,560 hours of development, resulting in more than 5,000 submissions.

The competition, co‑hosted by Sohu and Tsinghua University’s Computer Science Department, raised the difficulty by requiring both entity recognition and sentiment analysis.

In the final showdown in Beijing, 10 teams advanced to the algorithm summit, with the champion team from Beijing University of Posts and Telecommunications receiving a ¥40,000 prize.

Other top teams included Dalian University of Technology + University of Electronic Science and Technology of China (¥10,000), Suzhou University (¥10,000), and joint teams from Jinan University + Huazhong University of Science and Technology and others, each earning ¥5,000.

Charles, Chairman and CEO of Sohu Group, praised the participants as “fresh minds” and encouraged more talent to join Sohu’s AI initiatives.

Post‑competition reflections highlighted three key lessons: (1) focusing on the data itself can reduce noise and accelerate model iteration; (2) model ensemble strength depends on a sound fusion strategy rather than sheer quantity; and (3) solid engineering practice, including bug fixing, is crucial for performance gains.

Beyond the technical challenges, the event featured various activities such as rock‑climbing, tea breaks, and a banquet, emphasizing community building alongside algorithmic advancement.

The organizers expressed hope that the competition will continue to inspire breakthroughs in natural‑language processing and invite more algorithmic talent to solve future challenges.