Retraction Retraction: Research on Personalized Healthy Diet Recommendation Based on Artificial Intelligence ( J. Phys.: Conf. Ser. 1852 022085 )

. With the development of the Internet, China's good data mining technology in e-commerce has achieved development, and the recommendation service has gradually become popular and has been adopted by the majority of citizens. In order to facilitate people to quickly find what they need from a large number of commodities and save consumers' time, the recommendation system pays more and more attention to the user experience. This paper constructs a healthy diet recommendation model based on improved Apriori. When judging the user's dietary preferences, the improved Apriori algorithm is used to recommend the diet. The original algorithm and the improved algorithm are compared to the same example. The results show that the improved algorithm has advantages in mining association rules and can better realize personalized healthy diet recommendations.


Introduction
With the rapid development of my country's economy, people's living pace is getting faster and faster while people's living standards are improving. People's poor living habits have caused a series of health problems. Unreasonable eating habits and dietary structure will affect people's health. And how to develop healthy eating habits and ways in busy work is a problem that people are concerned about. Therefore, people have gradually noticed the importance of food choices for three meals a day, and a series of nutritional products have been derived. However, due to the large population of our country and the lack of talents in the nutritionist industry, not everyone can have a professional dietitian to arrange the corresponding food intake according to personal physique. The professional dietitian is expensive and inefficient, the penetration rate is not high, so the intelligent diet recommendation designed by the computer is gradually adopted. This article proposes a diet recommendation model based on improved Apriori diet recommendation model, and analyzes its effect. Apriori algorithm is a classical frequent itemset algorithm for mining association rules. It mainly finds out the relationship between itemsets in the database through the iterative way of layer by layer search, so as to form rules. The process is mainly composed of connection (class matrix operation) and pruning (removing the unintended intermediate results). The basic idea is: firstly, all candidate 1-itemsets C1 are found, and then frequent 1-itemsets L1 are generated according to item set C1. The frequency of occurrence of these frequent sets needs to be equal to or greater than the given minimum support, that is, association rules satisfying the minimum confidence and support are generated from frequent sets. Secondly, candidate 2-itemsets C2 are generated by L1 according to iteration rules, and frequent 2-items are generated from C2 by C2 Set L2, in turn, until the location of the frequent itemsets can no longer be generated. The circulation flow chart is as follows:

Apriori Algorithm Improvement
As a classic association rule mining algorithm, Apriori algorithm has a great influence in the development and research of data mining technology. The algorithm can mine data through repeated connection operations and pruning operations. However, with the increase of the data mined, the defects of this algorithm are gradually revealed, and its main disadvantages are as follows: (1)When the number of candidate itemsets is large, the time consumed by the algorithm will increase accordingly. When a transaction database contains equal to or greater than 100 frequent 1-itemsets, the number of candidate frequent 2-itemsets generated is 200. The calculation of such a large number of candidate sets will consume huge memory space and very long time.
(2)The database will be rescanned along with the connection operation. Multiple scans means that multiple input and output operations are required. In the long run, the efficiency of the algorithm will (3) Once the support of the algorithm is determined, it cannot be changed, otherwise it will affect the accuracy of data mining.
This study improves the Apriori algorithm from the aspects of reducing the generation of scanning transaction databases and frequent itemsets, so as to achieve the purpose of optimizing the Apriori algorithm and improving its computational efficiency. The specific improvements are as follows: (1) Scan the candidate set C1 generated by the transaction database D, and calculate the support of each candidate set.
(2) Analyze the candidate item set C1 to obtain L1.
(3) Analyze L1 and find the most supported items, and list the possible candidate item sets one by one according to the obtained support item set (k-item set Ck(2,3,... )).
(4) Analyze the support of each data in the frequent K-items set Ck, and then use the properties of the Apriori algorithm to compare and calculate the obtained data one by one, and finally get the items set Cn.

Instance Data
Use the Apriori algorithm to mine the frequent itemsets of the data in the system constructed in this article. The following table shows the dishes in a user's attention list in the system:

Mining of Association Rules before Algorithm Improvement
From the diet menu of a user in Table 1, it can be learned that there are 10 diet records related to the user, then the value of D is 10 and the steps of using Apriori algorithm to mine frequent itemsets are as follows: (1) During the first round of the algorithm, each item that meets the standard conditions is an element in the candidate 1-item set C1. The algorithm is used to scan D in the database to determine each item in C1 The support of the element, as shown in Figure 2:   Figure 3. Search candidate 2-item sets and frequent 2-item sets (4)Therefore, the content in the frequent 2-item set L2 is determined, and its generation is composed of all 2-item sets with support greater than the minimum support in the candidate 2-item set C2.
(5) Use the algorithm 1 1 L L ⊕ to calculate the content in the frequent 2-item set L2 to obtain the candidate 3-item set C3. The calculation process is as follows: i. Connection process: i. Connection process: 9  7  9  3  8  3  7  3  9  2  7  2  9  1  8  1  7  1  6  1  3  1  2  2  3   ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I 7  2  1  9  7  9  3  7  3  9  2  7  2  9  1  8  1  7  1  6  1  3  1   I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I 9  8  1  9  7  1  8  7  1  9  6  1  8  6  1  7  6  1  9  3  1  8  3  1  7  3  1  6  3  1   ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  ,  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I   {  } {  }}   9  7  3  9  7  2   ,  ,  ,  ,  ,  I  I  I  I  I  I ii. Use the properties of the Apriori algorithm to perform pruning operations, and then define all the subsets in the frequent itemsets as frequent itemsets according to property 1. The following is the process of judging whether the candidate set contains infrequent itemsets: In the first step, the subsets of { }

Analysis of computational efficiency before and after the improvement of Apriori algorithm
Through the above examples, it is not difficult to see that the biggest difference between the improved Apriori algorithm and the pre-improved Apriori algorithm is that the improved Apriori algorithm only needs to scan the transaction database twice, and it is analyzed from the structure of the first scan The candidate k-item set Ck (k>2) that may exist in the database is generated, thereby reducing the generation of candidate sets. In order to test the difference in the operational efficiency of the Apriori algorithm before and after the improvement, different amounts of data were used for testing, and the results are shown in Table 3:

Conclusion
With the development of Internet technology in China, various recommendation systems have been sought after by people. This article builds a framework for a healthy diet recommendation system based on artificial intelligence, which is an intelligent recommendation system that recommends different dishes according to different users' physical conditions and dietary preferences. In order to build a recommendation system with a better user experience, a healthy diet recommendation based on the Apriori algorithm is constructed, thereby reducing the number of database scans, achieving the purpose of reducing the generation of candidate sets and improving the computational efficiency of the algorithm.