Automatic Decision Algorithm of Interpretation Power in Criminal Justice Based on Data Activity Consultant

In order to solve the problems existing in the current decision algorithm, such as poor data processing performance, low decision accuracy, and long decision time, an automatic decision algorithm of criminal justice interpretation right based on data activity consultant was designed. According to the requirements of reasonable design consultant system data activity and demanded data activity consultants provide data processing requirements and scope. Use the Scrapy web crawler framework to crawl data related to criminal justice interpretation and criminal law provisions from related websites, and clear and extract the collected data to realize data query. Based on the obtained data, the feature array of criminal law is designed, and the decision of criminal judicial interpretation right is made. The C4.5 decision tree algorithm is used to predict the correct rate of decision. The decision of criminal judicial interpretation right is adjusted constantly according to the prediction results to achieve the goal of the automatic decision of criminal judicial interpretation right. Experimental results show that the algorithm has superior data processing performance, high decision accuracy, and short decision time, which veriﬁes the eﬀectiveness of the algorithm.


Introduction
e personal data activity consultant system was first born in Germany. In accordance with Article 4f, Paragraph 1, of the Federal Data Protection Act [1], public institutions and private law subjects (except private companies with less than 9 persons) automatically collect, process, and use personal data or, otherwise, personal process data and have at least 20 permanent employees. Private subjects that conduct automated data processing for preinspection, transmit or anonymize personal data in business, or automate data processing for market or research purposes must appoint, in writing, a permanent consultant for personal data activities. If a personal data activity adviser is not appointed or does not do so within a prescribed period or in a prescribed manner, or if an appointed personal data activity adviser is unqualified or untrustworthy, the enterprise is liable for an administrative fine of not more than €50,000.
Data activity consultants should have general knowledge of data protection and the necessary knowledge relevant to the specific data activities of the enterprise involved. Although the criminal law only requires personal data activity consultants to have the necessary skills and does not require specific vocational training, in practice, it is generally required that personal data activity consultants have a "technical background." Personal data activity consultants are highly professional, so they can read the algorithms directly and find problems more easily than ordinary people and administrative officials. e personal data consultants themselves are employees of the enterprise, and the business secrets of the enterprise are protected by the trust relationship between the enterprise and the personal data consultants. e administration has also been appropriately modest, thanks to the presence of consultants on personal data activities. In Germany, personal data activity consultants have always played a major role in regulating corporate data activities. By emphasizing the principle of self-control over data activities, the political system strives to avoid interfering with the enterprise's day-to-day economic activities and maximize respect for corporate freedom.
Judicial interpretation of criminal law is a kind of authoritative interpretation with the effect of criminal law, and it is the highest level of trial in China. In the specific application of criminal law, procuratorial organs directly clarify the connotation and extension of criminal law norms that procuratorial organs should follow at all levels of trials in the country [2]. erefore, judicial interpretation of criminal law is of great significance for law enforcement personnel and citizens to correctly understand the exact meaning and scope of application of criminal law norms. It is also one of the important means to materialize and finally realize the legislative intention reflected and embodied by criminal law norms [3]. So this article will discuss data activity consultant where this system is applied to the decision-making of criminal judicial interpretation to make political system prevent the risks of large-scale goals to achieve. Especially given the activities related to the criminal justice and some reasonable explanation and suggestion [4], this would also help ease the pressure of the administrative organ, avoid administration bureaucracy ills, and guarantee social stability [5]. However, most of the current decision algorithms are based on artificial intelligence [6]. However, it is found in the practical application that this algorithm has such problems as poor data processing performance, low decision accuracy, and long decision time, and its practical application effect is not good. erefore, this paper mainly carries out the design of automatic decision algorithm of criminal judicial interpretation right with data activity consultant as the core, in order to improve the professionalism and correctness of criminal judicial interpretation right decision and promote the further development of enterprise and society in a safe and stable way.

Design of Automatic Decision Algorithm of Criminal Judicial Interpretation Power
In the following sections, we provide details about the decision algorithm of criminal judiciary interpretation.

Criminal Justice Interpretation Power Data Processing
Based on Data Activity Consultant. In order to improve the accuracy of automatic decision-making of criminal judicial interpretation power, it is necessary to design a reasonable data activity consultant system. erefore, this paper focuses on the design of the system, and the specific requirements are as follows: (1) e data activity consultant shall be professional and reliable and meet the qualification requirements specified in the criminal law. (2) If it fails to perform this obligation, it shall be subject to administrative punishment. (3) e data activity consultant shall have the obligation of confidentiality and shall not disclose the trade secrets obtained due to his position. (4) e data activity consultant shall actively impart data activity compliance knowledge to relevant personnel.
(5) e data activity consultant shall participate in the design, inspection, and evaluation of algorithms and write and save algorithm reports. (6) Data activity consultants shall regularly collect criminal justice information. (7) e data activity in this paper effectively creates multisystem connection points for an in-depth consultant shall actively cooperate with the criminal law department. (8) e data activity consultant has the right to seek help and support from the data activity regulatory authority. (9) If the data activity consultant neglects performing his duties or commits an illegal act jointly with relevant departments, he shall bear joint and several liabilities.
erefore, through the data activity consultant system, the internal criminal justice organs should be understood to maximize the compliance of data activities without compromising the effectiveness of criminal justice. Today, with the wide application of algorithms and the rapid development of machine learning, data activity consultants can give full play to the role of the whole data collection process of criminal justice interpretation power. erefore, data activity consultants must understand the data processing program and be notified in time before a new algorithm is implemented. Only in this way can they get enough time to understand and make correct decisions on the power of criminal judicial interpretation. is paper takes the data activity consultant as the core to collect and process the data of the power of criminal judicial interpretation to lay a solid foundation for the follow-up algorithm research. Before collecting and processing data on the right of criminal justice interpretation, design a reasonable data activity consultant system according to relevant requirements, ask the data activity consultant to give relevant data processing requirements and scope, use the Scrapy web crawler framework to crawl criminal law provisions and other data related to the right of criminal justice interpretation from relevant websites, and clean up and extract the original data and realize the data query and display of criminal judicial interpretation power. e data processing architecture of criminal judicial interpretation power consists of three modules: data acquisition, data processing, and query display. e basic framework of the data processing architecture is shown in Figure 1. erefore, the target data to be collected includes various criminal law provisions. e data acquisition module uses the Scrapy framework and Selenium testing tool to access relevant websites, extract text data in websites, and store it in the MongoDB database for subsequent processing after Bloom filter removal [7]. is paper introduces the implementation of Scrapy crawler through the criminal law of a website. First, use Selenium and PhantomJS test tools to access the site's home page, simulate the browser for rendering, and directly grab the rendered page. en, BeautifulSoup is used to find the navigation tag for the feedback response information, obtain the starting URL address of various criminal provisions, and carry out the deep-first search strategy to crawl the data of criminal provisions [8]. rough observation, it is found that the tag name of each criminal law article is caseName. By obtaining the hyperlink address information in the tag one by one, the URL address of each criminal law article can be obtained by concatenation with the previous category address. At this time, the Bloom filter is used for URL deprocessing, and the target URL is passed into the Bloom filter to check whether the URL exists in the collection [9]. If the URL address already exists in the collection, it will be abandoned; otherwise, it will be queued and the address after hash mapping will be added by 1 [10]. Finally, the URL address is read from the Redis request queue to enter the interface of the body of the criminal law articles. Relevant information such as the title, body content, and execution number of the criminal law articles is obtained. e do-cld attribute value in the request is taken as the unique identifier of the criminal law articles and stored in the MongoDB database [11].

Data Processing Module.
As the data obtained by the crawler is rough and unfavorable for subsequent retrieval and analysis, the data processing module is adopted to extract and process the original data collected before and extract the key information from the text. By analyzing the general format of criminal law provisions, appropriate matching template is designed to extract key information and store it in MySQL database. In addition, Jieba word segmentation and Term Frequency-Inverse Document Frequency (TF-IDF) operation are carried out on the contents of criminal law articles to extract the keywords of each criminal law article to facilitate subsequent retrieval and analysis.
In the text keyword extraction part, Jieba word segmentation [12] and TF-IDF technology are adopted. Jieba has three word segmentation modes: precise mode, full mode, and search engine mode. In addition, it can tag the word after word segmentation. Since the precise mode will cut the sentence accurately, which is suitable for text analysis, this paper adopts this mode for word segmentation [13]. In order to improve the accuracy of words, through the research and analysis of massive criminal law provisions, this paper stores customized keywords in Jieba user dictionary and sets the part of speech as nz. In the provisions of criminal law, nouns and verbs can better reflect the content of criminal law provisions. erefore, the parts of speech after text segmentation are extracted from n (noun), v (verb), and nz (custom word) and imported into the bag of words model; the occurrence frequency of keywords is calculated, and the words whose frequency is less than 5 times are eliminated, so as to reduce the influence of incorrect word segmentation and other factors on subsequent processing. en the TF-IDF operation is carried out on the extracted keywords of criminal provisions. At the same time, in order to improve the accuracy of keyword extraction and overcome the defect that TF-IDF cannot reflect the position information of words, this paper subsections the content of criminal provisions and strengthens the key paragraphs [14].
Tf-IDF is a weight calculation method used to evaluate the importance of words after word segmentation in the text. It is commonly used in fields such as paper mining and keyword extraction [15]. e common formula is as follows: Among them, N W is the number of occurrences of keyword W in criminal law articles, and N is the total number of words after word segmentation in criminal law articles. Y is the total number of criminal law articles, and Y W is the number of criminal law articles including keyword W. It is not difficult to see from the formula that although common words appear frequently in the provisions of criminal law, i.e., N W is a large number, there are also many  provisions of criminal law containing this word, i.e., Y W is also a large number, resulting in the frequency of words in inverse documents being close to 0 and their weight being reduced [16]. us, it can be seen that TF-IDF tends to filter out common words and retain important words. After calculating the TF-IDF value of the words in each article of criminal law, the five words with the highest weight value are stored in the database as the keywords representing the criminal law provisions for the convenience of subsequent query [17].

Querying the Display Module
(1) Retrieval Display Function Implementation. e data retrieval part is based on the SQLAlchemy framework, which provides an SQL toolkit and object-relational mapping tools. Developers can carry on persistent operation to the database in the way of object oriented, which improves the efficiency of development. e specific use steps are as follows: (1) Design the required entities in the model file. e entity class inherits the instance of the base class and the field names and attributes are consistent with the corresponding MySQL page. e client passes the retrieval field to the appropriate router path address through Okhttp, and SQLAlchemy matches the request field and returns the result to the client. In addition, in order to ensure the user experience and avoid the waste of resources, the Paging module is programmed to perform Paging operations on the data of criminal law provisions and improve the efficiency of system use. When the user rolls the list near the bottom, the background automatically sends a data request to the server to load new data and update the current interface. e request time is short and the interface updates quickly, which improves the robustness of the system. After clicking on the title of a criminal law article, the page will jump to the details page of the criminal law article to view the specific content of the criminal law article. In addition, click the summary button in the upper right corner of the details page to trigger the ToolBar navigation bar, showing the key information of the criminal law article.
(2) Implementation of Recommendation Function. When using the query function, the user may not directly enter the query criteria but need similar results. If there is a recommendation function to provide users with some criminal law provisions related to query conditions, it is very convenient for users. Recommendation function is based on Word2Vec, using TF-IDF technology to weight word vector. It strengthens the influence of keywords in criminal law articles, effectively uses word vector to express criminal law articles, calculates the distance of word vector of criminal law articles by cosine similarity formula, and pushes the most similar to the first five criminal law articles. e process is divided into two steps, as follows: e first step is to generate a vector model of criminal law articles: firstly, the collected criminal law articles are preprocessed by word segmentation to remove symbols and numbers, and relevant part-of-speech words (nouns, verbs, and custom words) are extracted by pos tagging. en, the training samples were imported into Word2Vec, and the word vector of each word was obtained by using Skip-Gram model for training. en, each article of criminal law is divided into words, and the word vector of each word and its TF-IDF word frequency weight value are weighted average, so as to get the document vector and index list of criminal law articles and generate the vector model. e second step is to calculate the cosine similarity: after the display interface into the criminal law on the page, the provisions of the criminal law will be all documents in the document vector and the vector model vector of the cosine similarity computing, and the similarity of results to sort and take out the top five articles from the index list with related law is recommended to the user interface.
Among them, A and B represent different criminal law article vectors, and A i and B i represent the components of vectors A and B.

Design of Automatic Decision Algorithm of Criminal
At present, the big data reserve of criminal law cases and criminal law relationship database is obviously insufficient in China, and there is also a lack of corresponding corpus for word segmentation and integration of criminal law language. erefore, if the intelligent decision-making function by means of machine learning is needed, it can only rely on manual sorting of corresponding database samples. To solve this problem, a training database for a specific class of relationships needs to be established. e field and record structure of the database is shown in Table 1.
e total number of recorded lines is D, the above standard k of each line identifies the specific criminal law situation contained in the data record, the name of the field Req  Table 1.
In the value of each field, because the characteristic attribute that constitutes a specific criminal law relationship is usually only the dimorphism of "being" and "nothing," the value of field Req[n − 1] in each record line can be denoted as "1" or "0" correspondingly. Similarly, the value of field Jud is usually 1 or 0 to indicate whether it is correct. As for the value of field Mode, since this field does not need to participate in specific datalization operation, and this field is essentially the observation result of the record row, its value can be any type of data. In order to facilitate the presentation of this paper, in the Mode fields, the continuous natural numbers starting from 1 refer to the relation model of each subclass of criminal law, so as to form a specific array of Mode[m − 1], so as to make the preliminary decision on the right of criminal judicial interpretation.
In order to realize the prediction of the correct rate of the decision of the right of interpretation of criminal justice, the first problem to be solved is how to summarize the scattered criminal law data into a specific criminal law relationship model. In this regard, this paper suggests that a complete criminal law relationship database should be established according to the above-mentioned data sample set, and then a certain criminal law data can be classified by using the classification algorithm of data science. erefore, the core of decision tree classification algorithm is to select the most appropriate feature attribute as the branch node, so that it can predict the classification of each sample more accurately, and the size of the tree is as small as possible, which can establish a reasonable connection with the goal of the decision of criminal judicial interpretation right and become the most appropriate classification algorithm. e comparison of three different mainstream decision trees is shown in Table 2.
In this paper, referring to the classical algorithm of C4.5, the classification of criminal law situation is appropriately improved. e following three steps are used to realize the classification of criminal law relationship model.  Table 1 After the combined record lines, the split information entropy of the feature attributes of each constitutive element can be calculated according to the   (Req[t]) can intuitively show the complexity of a certain characteristic attribute. e higher the value of entropy is, the more unstable the corresponding constitutive requirements of criminal law in the ideal state will be, and the more feasible it will be to change to other similar constitutive requirements.
(2) Compute the information gain of characteristic values Information gain is another metric for data attributes that is different from information entropy in the basic theory of data science. e more important the influence of an attribute on the observation results is, the more the uncertainty space for classification will be eliminated by it, that is, the more the information for classification will be provided. From the perspective of the fuzziness of criminal law relations, if a certain constitutive element Req[t] is more important to the determination of criminal law relations, the form of its corresponding element attribute in the data sample set is less complex, that is, the information gain is greater, and it is easier to determine it in the decision-making process of criminal judicial interpretation power. erefore, the following formula for calculating the gain of component element G � (Mode, Req[t]) is proposed in this paper, which is analogous to the calculation method of information gain in data science According to the above form of information gain rate collection, select one of the biggest values of Max(G R [n − 1]) as labels for a particular data set for the first time classification properties; thus the binary or fork forms a number of data sets, again on the basis of the above three steps of the recursive classification, screening each subsample of the maximum of information gain rate points to the attributes. Use this as a tag attribute for subclassing. Finally, the smallest subclass obtained by recursion has the same observation result of the criminal law relation model; that is, it Mobile Information Systems conforms to the algorithm objective of constructing the tree to determine the classification standard, and the label of each branch node is clear. erefore, the construction of criminal law classification decision tree is completed by learning and training according to the existing data samples, and then the unknown results of the criminal justice interpretation right facts can be included or classified forecast according to this tree model. Naive Bayes is a classification method based on the mathematical Bayes' theorem and the independent assumption of characteristic attributes on the corresponding data. e data conclusions obtained by naive Bayes algorithm are generally presented in the form of probability. In the context of criminal law application, naive Bayesian algorithm can be used to predict or analyze the risk of the correct rate of criminal judicial interpretation right decision based on historical data. e realization process is to learn and train the data samples and conclude the specific probability model according to the rules embodied in the samples. To realize the prediction of the probability conclusion of specific unknown facts according to the model and application model, based on the data table structure model established in Table 1, the improved application of naive Bayes algorithm in criminal law scenes and the realization of the corresponding automatic decision algorithm of criminal judicial interpretation right are explained as follows.
Under the naive Bayes algorithm of probability model is actually according to the observations in the training sample prior probability forecast a posteriori probability but needs to assume that the prior probability calculation process conditions are independent of each other and do not have relationship between the cross, but the premise is basic scenario of criminal law that can be in accordance with it; the constitutive requirements of a scientific theory of the basic criminal law are independent elements. erefore, under this premise, the calculation formula of prior probability can be deduced from the combination of known facts feature elements R[n − 1] (k) for the unknown disputes that conform to all the features of constitutive elements in line k under the classification of inclusion and absorption, where P(Jud � 1) represents the accuracy of observation results recorded by all data, and R(Req[n − 1]) � R[n − 1] (k) |Jud � 1 represents the corresponding accuracy of each feature under the condition of matching the value of each feature in the feature set of facts.
In addition, it is also necessary to calculate the total probability R(Req[n − 1] � R[n − 1] (k) ) for the value of each feature in the factual feature set, that is, the sum of the correct rate and error rate for the value of each conditional feature. e calculation formula is as follows: us according to Bayes' theorem, from the results of the prior probability and the probability to deduce specific criminal law relations under the model of criminal judicial interpretation accuracy probability model, the probability model of the practical significance is the historical decision results and the unknown criminal judicial interpretation Mobile Information Systems 7 decision-making process; the results established a degree of connection, making the former to the latter provides a reasonable prediction.
As shown in the above equation, the posterior probability obtained from the combination of certain characteristic elements by this equation is the correct rate of predicting the decision of criminal judicial interpretation power according to the constitutive elements of criminal law relations formed by it. When the probability of correct rate is higher than 0.9, it is generally considered that the decision of criminal judicial interpretation power is high and the result can be directly output. However, if the probability is lower than 0.9, it is necessary to constantly adjust the decision result, so as to achieve the goal of improving the correct rate of automatic decision of criminal judicial interpretation power.

Experimental Design.
In order to verify the effectiveness of the automatic decision algorithm of criminal judicial interpretation power based on data activity consultant, an experiment was designed. e specific experimental environment is shown in Table 3.
Due to the complexity of criminal law relationship, the number of constituent elements is relatively large. is paper takes a simple criminal law relationship with 7 elements as an example and puts all of it into the data sample and participates in the algorithm implementation. On this basis, all judicial interpretation information related to the provisions of the criminal law will be collected, and the collected data will be cleaned and processed, which will be taken as experimental sample data, so as to verify the judgment algorithm proposed in this paper.

Analysis of Experimental Results.
In order to verify the effectiveness of the designed automatic decision algorithm of criminal judicial interpretation power with data activity consultant as the core, the time of data collection, processing, and query under different data levels is first tested.
e specific results are shown in Table 4.
With the increase of amount of experimental data, this paper's algorithm of data acquisition, processing, and query time tends to rise, and under the condition of the experimental data of 500 GB, the algorithm of data acquisition, processing, and query time reached the maximum value, a maximum time of 1.37 s of data collection, and data processing time to a maximum of 1.42 s. e data query time is 0.61 s, and the data collection, processing, and query time of this algorithm are low, which meets the research expectations. On the above basis, the decision accuracy of the algorithm in this paper is compared with the automatic decision algorithm based on artificial intelligence. e decision accuracy refers to the degree of similarity between the decision results of different methods and the decision results of authorities. e specific comparison results are shown in Table 5.
By analyzing the data in Table 5, we can see that the accuracy of automatic decision-making of criminal judicial interpretation power of this algorithm is 96.9% at the highest and 93.8% at the lowest. e accuracy of automatic decisionmaking of criminal judicial interpretation power based on artificial intelligence algorithm is 86.5% at the highest and 71.9% at the lowest. Compared with the automatic decisionmaking algorithm based on artificial intelligence, the automatic decision-making accuracy of criminal judicial interpretation power of this algorithm is higher, which can lay a solid foundation for the analysis and research of criminal judicial interpretation power.
Finally, the decision-making time of this algorithm and the automatic decision-making algorithm based on artificial intelligence is compared, and the comparison results are shown in Figure 2.
By analyzing the data in Figure 2, it can be seen that the automatic decision-making time of criminal judicial shorter. e overall operation efficiency is higher, and the automatic decision-making of criminal judicial interpretation power can be realized faster.

Conclusion
e automatic decision algorithm of criminal judicial interpretation power is not only a criminal law problem, but    also a political and economic problem. e political system needs to prevent social unrest caused by large-scale harm. erefore, it is of great significance to design the automatic decision-making algorithm of criminal judicial interpretation power. erefore, in order to solve the problems existing in the current algorithm, such as poor data processing performance, low decision accuracy, and long decision-making time, this paper designs an automatic decision-making algorithm of criminal judicial interpretation power with data activity consultant as the core. According to the requirements, the data activity consultant gives the relevant data processing requirements and scope, collects the criminal law provisions and other data by using the Scrapy web crawler framework, processes the collected data, and provides the data query and display function, so as to complete the design of the data processing architecture. On this basis, the array of criminal law characteristics is designed, and the decision of criminal judicial interpretation power is made. C4.5 decision tree algorithm is used to predict the accuracy of decision-making, and the decisionmaking of criminal judicial interpretation power is continuously adjusted according to the prediction results, so as to achieve the goal of automatic decision-making of criminal judicial interpretation power. e experimental results show that the maximum data acquisition time of this algorithm is 1.37 s, the maximum data processing time is 1.42 s, and the data query time is 0.61 s. e data acquisition, processing, and query time are low. e highest accuracy of automatic decision-making of criminal judicial interpretation power is 96.9%, the lowest is 93.8%, and the decision-making time varies from 0 s to 0.65 s. It can realize the correct and rapid decision-making of the power of criminal judicial interpretation and can provide important theoretical reference for relevant research.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.