Generally speaking, the difficulty in creating a big data legal supervision model does not lie in technology, but in identifying abnormal supervision points and obtaining data. Among them, data acquisition needs to first rely on abnormal supervision points to determine the data range. Therefore, identifying abnormal supervision points plays a fundamental role in creating a big data legal supervision model. In judicial practice, the identification of abnormal supervision points is summarized and refined by prosecutors in handling cases and carrying out special activities, which is the inherent meaning of "business leadership" proposed in the digital procuratorial work mechanism. However, in practice, there are also directional errors in the understanding of abnormal supervision points by some prosecutors, who do not start from the functional positioning of legal supervision agencies, resulting in an awkward situation of "plowing others' fields and wasting their own land". Accurately identifying the supervision points of the big data legal supervision model is to identify the universal "small incisions" that the big data legal supervision model relies on. At present, we should mainly focus on the hot topics of serving the overall situation, prominent issues in law enforcement and justice, weak areas of social governance, and the absence of public interest protection. The author starts with mature big data models used in various regions and finds that their supervision points mainly focus on prominent issues of administrative law enforcement and judicial authorities, including the following four "small incisions". One is the illegal performance of duties. One is the abuse of power by administrative and judicial personnel, who use their power to seek rental space and seek illegitimate benefits for others. For example, the supervision model for drug abusers evading compulsory drug rehabilitation cases aims to address the issue of individual police officers helping drug abusers forge minor criminal cases, which result in drug abusers being sentenced to a short sentence of less than one year and thus avoiding being forcibly isolated for two years for drug rehabilitation. In this regard, prosecutors can collect minor criminal cases where drug abusers surrendered before and after being forcibly isolated for drug rehabilitation or during the period of compulsory isolation for drug rehabilitation, and analyze the data, Lock in police officers who help forge fake cases. Another type is that administrative and judicial personnel should act but fail to perform their duties in accordance with the law. The downward supervision model for minor injury criminal cases is mainly aimed at some minor injury cases where the public security organs do not conduct injury assessment on the victim and directly impose administrative penalties on the suspect. The procuratorial organs can access the administrative penalty data of the public security organs, screen out the data that has not been assessed, and further identify cases where the victim's injury has already met the level of minor injury or above, So as to determine the supervisory clue of "using punishment instead of punishment". The second is to cover up illegal incisions in a legal form. One approach is to look at the data of a particular case alone, and there are no issues with case processing. However, once a large amount of data is collected and expanded, illegal purposes will be discovered. For example, in the supervision model of vehicle insurance fraud cases, looking at a single case, the evidence of the case can prove that the relevant vehicle has had an accident, and after being repaired by a repair shop, it claims compensation from the relevant insurance company to obtain relevant profits. However, if a large amount of data similar to judgments is retrieved, it may be found that a large number of lawsuits may be related to a repair shop, which may be suspected of some clues to civil legal supervision cases, Furthermore, it may lead to clues to criminal legal supervision cases such as suspected fraud. Another reason is that there are data barriers between institutions, resulting in legitimate forms of concealment of illegal purposes