This article is reproduced from Netease Journalism College On August 22nd, Netease News “Zhong Xueshe Salon†took the theme of “How Artificial Intelligence Can Reshape Journalism†and invited experts and scholars from Peking University, Zhongda University, Reuters, and Xinhua News Agency to discuss the impact of new technologies. Under the media reshaping. The following is the reorganization of Wang Haiming’s keynote speech by Reuters’ market development manager. From monitoring to verification, artificial intelligence identifies false news for reporters Reuters News Tracer is a tool for monitoring social media that was first used inside Reuters to help journalists and editors monitor news sources that appear in social media. Later, with the deepening of the algorithm, we found that using machine learning can also do more things. For example, for information sent out on social media, we can use algorithms to distinguish between news and general conversations. After three to four years of continuous optimization, this project is now gradually testing and opening up to external users. News Tracer was able to analyze all the content on Twitter that might be considered news: who sent the message first; who sent it first; and how trustworthy it was. If the analysis results determine the authenticity of the message, the corresponding 0% to 100% confidence value will be marked above. Reuters News Tracer interface Some time ago I happened to witness the spread of a public company’s rumors from publication to outbreak: When this false news was just released, I monitored it with the News Tracer tool. If I read the news source, I know it is false. It slowly grew to more than 800 hits, and was then republished by the Big V or other media. When it was reproduced by the website, it became news fermentation, but it was clarified in less than 24 hours. "I have witnessed the whole process because I received the first news source very early. The entire false news dissemination chain was initially limited to a small area and broke out with the retransmission of news websites and big V forwarding." The current application of News Tracer has the following aspects: 1. Automatically verify the true and false news, it will use the algorithm to analyze the issues that the reporter may usually pay attention to, to verify the authenticity of the news. 2. Remove unnecessary points such as advertisements, junk, rumors, and general conversations through algorithms and machine learning. According to different data sources to determine how true it is, at the same time it can remove noise, and will put the same category stacked together as a data group, while showing future updates. 3. Conduct real-time monitoring to verify news on Twitter and other social media. 4. With a large back-end database, customers can search for relevant news according to their needs and distinguish between true and false. Traditionally, artificial intelligence applications are distributed on that side, and Reuters News Tracer is a big data, artificial intelligence, machine learning application from the news source end. Reuters News Tracer can directly capture the information posted by the witness via social media and publish directly if the information is true. For example, after the earthquake in Japan, the earliest news on the Internet may be 4 minutes earlier than the first media release. Reuters News Tracer can conduct cross-validation as soon as it is published. If it is verified to be true, it will be released and it will be marked at the time of publication. What is the credibility of the news, which is 50%, 70% or 100%, and will show the source of the data for you to identify. 2 Machine Readable News to help users place orders in real time Since 2009, we have launched Machine Readable News products. The user's ordering system docks with our Machine Readable News to place orders automatically. This product has been innovated for several generations until now and is relatively mature. By automatically reading and analyzing news from the machine, the system can determine whether the incident is positive or negative, and then compares the results with the historical database, such as: how high the incident is positive, which companies are involved, and how good the news is. Judging the impact of news on commodities and stock prices. The real-time data and analysis obtained through the above method can allow the machine to identify the trading signals and place orders directly. Further, this product can also analyze market sentiment based on news, that is, give an index indicator to track the market's mood changes in asset classes or stocks. For example, commodities or gold, according to news reports during this period of time, is everyone's mood high or low? Our front-end has visualized the results and users can clearly see the results. The other is to give some economic indicators, economic indicators are relatively simple than the news, especially the stock market, the foreign exchange market, and now many stocks, foreign exchange, commodity market transactions come from direct orders for such indicators. 3 Establishing an enterprise relationship map through semantic analysis What value can be found in the news? Reuters has established a supply chain of corporate relationships through analysis of news. There are various networks of relationships among people, and so do companies. We have an engine that can analyze some company news. After analyzing, we can see that other companies mentioned in these news are its customers, competitors, parent companies, consumers, or subsidiaries. Make them a relationship. 犉 î¦î‚ˆ î¦î‚ˆ ┯ ┯ ┯ ┯ Here will use the Reuters accumulated products or data, 99% of the world's listed companies have ID in our database, it is easier to find those companies. We also have two key databases, one is an organizational database, the other is a human database, one is to do company associations, and the other is to do associations between listed company executives. After labeling, it is easy to separate upstream and downstream. connection relation. Third-party users who want to do data processing can directly use raw materials in the cloud to use our machine to generate associations. Comcn Electronics Limited , https://www.comencnspeaker.com