With the development of information technology and the wide application of big data, algorithm recommendation makes information dissemination more personalized, customized and intelligent, but there are also some chaos. It is reported that the personalized push of some news and information, online social networking and other platforms has the phenomenon of "more pan entertainment information, more vulgar content, more uncertified content" and "more than three", which also easily causes some users, especially teenagers, to indulge in the Internet. < / P > < p > with the help of algorithm recommendation, information acquisition has entered the era of "private customization" from "looking for a needle in a haystack", which brings convenience, but also has some negative effects. First, it will accelerate the formation of "information cocoon room" and "emotional contagion" effects, resulting in limited user vision. Second, it may lead to minors addicted to the Internet. The third is to provide convenience for the "big data kill". Once the "familiar" and "familiar" data become "familiar" users, it is very possible to judge whether the data is "frequently used" by users. The algorithm recommended by < / P > < p > should not be biased, but should be more "temperature". This requires designers and operators to take responsibility, relevant enterprises strictly abide by the law and discipline, and actively perform social responsibility. For example, we can establish a social evaluation mechanism to evaluate the consequences of using algorithms on the platform, so as to provide better services for users. At the same time, strong regulation is needed. Recently, the State Administration of Market Supervision issued the "anti monopoly guide on the field of platform economy (Draft)", which has made relevant provisions, and the relevant legislation needs to be further improved. Relevant functional departments should take action to strengthen the regulation of algorithm recommendation and punish suspected illegal behaviors according to law. In practice, algorithm recommendation should not be one-sided pursuit of efficiency, but also take into account public values, social ethics, etc., to ensure the algorithm recommendation on the right track. This requires regulators, platforms, enterprises and users to work together to make algorithm recommendation really benefit users.