没有合适的资源?快使用搜索试试~ 我知道了~
mathorcup数学建模挑战赛获奖论文-第四届B题_20025E.doc
1.该资源内容由用户上传,如若侵权请联系客服进行举报
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
版权申诉
0 下载量 66 浏览量
2024-03-14
22:08:30
上传
评论
收藏 242KB DOC 举报
温馨提示
试读
16页
mathorcup数学建模挑战赛获奖论文,历届,单项文件,内容丰富,大学生数学,数学竞赛,参考资料
资源推荐
资源详情
资源评论
The judges scoring, note
Team number:
20025
The judges scoring, note
The judges scoring, note
Problem:
B
The judges scoring, note
Title : Research about recommending books based on Hierarchical analysis method
and BP neural network
Abstract
At present, with the development of the information technologies and the Internet,
evaluation and ,the recommendation of all kinds of information is increasingly
concerned.
As far as the first question,the original data is filtered first.Then we can get Users’
evaluation on the books for 5 points. Considering the effect of different factors on the
books’ score,then the label, social friends, books browsing amount of three groups of
data were analyzed by bivariate correlation analysis respectively,so that we can get
the number of users on the books, scores and label users’ good friends, the history of
the book number of pageviews a positive correlation. Impact of users to book score
were history views > the number of users’ good friends > tag number of books.
As to the second question, This paper established the AHP model and BP neural
network model to predictthe score. The three influenced factor in grading is : The
number of books, bookmarks, browsing history, the amount of the user's friends ,
using which we establish evaluation index system. Then, through the establishment of
hierarchy analysis model, we get the label number, weight, number of users browsing
history friends three index.:0.0813,0.6837,0.2349. And then we determine the score
formula on the books of the user. Then 36 groups of data problems are analyzed, and
the use of scoring formula is obtained for each book user rating. Then we construct BP
neural network model: the original data were chosen for the first 99 sets of datas.We
input the number of books, different users label user number of friends,
booksbrowsing quantity and the corresponding book score,and we output the score
which is from the other users of the books. selecting 80 groups of training data for the
neural network,with the remaining 19 sets of data to test the model,making sure thehe
error is within 5.3%. Finally, using the network to predict user the trained on books
score. By comparing the two kinds of model so that we can obtain more accurate
results
As for question three, we consider that the higher the frequence of the comments
from the books , the higher user preference for books, through the users do not read
the book selection on the original data of the ID, select the top three of the score for
the books five points the highest frequency, which is recommended to the user's three
book by ID, then loop five times by five users will eventually get the recommended
books ID.
Key words:Books scores; correlation analysis; analytic hierarchy process; BP neural
network
- 1 -
Research about recommending books based on Hierarchical
analysis method and BP neural network
1. To description of the problem
The development of information technology and the internet has ended an
information-shortage era, but has brought the public overloaded information. In the
present era, both consumers and producers of information are facing significant
challenges: consumers find it is harder to have access to information they are
interested in, with explosively growing information around them; and producers also
find it is harder to make their information focused by readers.
The conflicts could be solved by a tool called ‘Recommendation’.
Recommendations are widely used in network products and Apps. Typical styles
include the relevant search, topics recommendations, products push in the
e-commercial, and ‘recommended friends’ search in various SNSs.
Attached data are the information of users’ behavior at a popular on-line bookstore,
including remarks on books, tags of books, and the users’ social networks. Consider
the following question with the attached data:
1.Analyze the effects on book remarks by users.
2. Design amodelto predict the remarks by users on their unread book.Information of
users is attached in the file ‘predict.txt’.
3.Recommedend every user (attached in the file;predict.txt’)three books,which they
have yet read before.
2. The analysis of the problem
Scores and book recommendation are mainly based on the amount of statistical data
processing. Therefore, we need to seize the key and useful date to solve the problem
and transform, screen, analyze, summarize the data to analyze the factors affecting the
user’s score to books, based on which we complete the user score prediction of books
and book recommendations. through the establishment of a model of user’s score of
books.
2.1 Analysis of problem 1
Problem 1 requires us to analyze the factors that affect scores of books. It is a
comprehensive analysis of the data in the annex. Firstly, we screen raw data of
user_book_score.txt and we get the data of 1-5 points by users; take into account the
different factors impact on scores of books, and then filter the data for other analyzes,
the relationship between the various stages of books initially been evaluated scores
and the number of labels, the relationship with social friends, relations with views of
books. Finally, the data obtained were analyzed and summarized the scientific obtain
affect users of books score factors.
2.2 Analysis of problem 2
- 2 -
Problem 2 requires to establish a model to predict user score for books in the annex
of predict.txt. First, the number of tags, social relationships, books to study three
aspects of views, which is a multi-objective decision-making problems. According
problem, you can use the software to build hierarchical analysis yaahp total score -
two-level model rule layer analysis, using integrated AHP to determine the total score
for each indicator weights, and determine a comprehensive evaluation formula books,
resulting in books score model for prediction score. Secondly, in order to more fully
consider the accuracy of the model, using BP neural network model, first create a
neural network structure, the number of different users to tag books, the number of
user's friends, books, book views and the corresponding score as input to predict other
users of books score as output, based on known data to train the neural network, the
process continuously adjust network structure until it reaches satisfied so far, and
finally the use of the trained network user score prediction for books.
2.3 Analysis of problem 3
Problem 3 requires us to recommend three unread books for each user according to
annex of predict.txt. Considering the book's praise higher the frequency, the higher
the level of the user's favorite books, using spss for user_book_score.txt attachment
filter, get a fifth of all books rated frequency, and then sort processing in EXCEL,
using LOOKUP function screening this user has not read the book to get the ID, select
three of the most frequently rated as fifth book is recommended to the user's three
books ID, and then finally get circulation problems were analyzed five times the
required five users Recommended books ID.
3. The provisions of the symbol
4. The assumption of the question
1.Assuming that the impact on the user rating factors are independent of books.
2. Assuming that the only factors affecting the score of books are the books labels,
the number of the user's friends and books views.
3.Assuming that the frequency of the book praise higher , the user love the book
more.
symbols
Explanation
N
the number of Observing samples
y
User ratings for books
1
x
User traffic indicators
2
x
The number of indicators users’ Friends
3
x
Tag number of indicators
CR
Consistency proportion
剩余15页未读,继续阅读
资源评论
阿拉伯梳子
- 粉丝: 1648
- 资源: 5735
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功