Book Recommendation System
We designed and developed a book recommendation system that recommends the next few books for our target users based on their own reading tastes.
R Shiny App
Eda Zhang, Radhika Kulkarni, Daye Kang, Eunhee Sung
Collaborative Filtering, Explorative Data Analysis, UI mockup
2.2020 - 5.2020
In this project, we designed and developed a book recommendation system that recommends the next few books for our target users based on their own reading tastes. Here is the current version app: https://edaxplor.shinyapps.io/book_v4/ The data source we used comes from the UCSD Book Graph website, and a Book Recommender Project from Kaggle. Considering the scope of our project, and our target users to be mainly children younger than 15-years-old, and educators of children at that age group, we mainly used two datasets from the websites: a book dataset of the fantasy and children’s books genre and a rating dataset with users ratings of the books. Logistic and Linear regression model is tested for the global test of model adequacy and showed some linear relationship between the variables.
To figure out the best algorithm for our product, we analyzed and compared the datasets with both, using unsupervised learning. From the results of the analysis, it showed that the unsupervised learning methods performed better than the supervised learning ones. Among the different supervised learning techniques we implemented, the linear model: fast and frugal tree outperforms other models including generalized linear model and the non-linear model xgboost. We included the details of our analysis in the model interpretation section. In terms of the unsupervised learning, we found that the collaborative filtering clustering method fit the goal of our project the most. Other clustering methods we implemented including: dimension reduction techniques with UMAP and K-means. Eventually, we used the UBCF algorithm to make recommendations to our users based on their own preferences of the books, and we captured the main idea of the book with the TF-IDF analysis on the book descriptions.
As a result, a book recommender is developed as a shiny app in R, and it is built with genre and rating filters. After users select their favorite genre and rate on books that they have read, the model gives a suggestion of 3 books. Also, as a reference, a text cloud would be driven to see which words are frequently using in the book.
It is not an easy job to recommend to others what book to read because people spend their money to buy a book and spend their time reading it. There are some helpers to suggest the next book that readers might like, such as best-selling books displayed for different genres in bookstores, and search and read other’s reviews online. It might help people to decide a book they might like, but it is not always true. Since every person has their own reading tastes, picking a book that they might like takes time. To save their time and make their decision easier, the book recommendation system will be introduced to propose an recommendation algorithm that can save them from all the pains of searching online, reading all the reviews, and comparing with their tastes.
In this project, a book recommendation algorithm is built with two genres, children’s book and fantasy book. Therefore, any person who would like to read a new book, especially fantasy lovers, children, their parents and educators who teach age under the age of 15-years-old are the targets for this application.
Data sources and Preprocessing
To develop the application, we used four datasets. The main data set used to develop the model is the Genre dataset . The dataset was collected in late 2017 from goodreads.com with updated several files in May 2019. It has 26 columns and 1,242 rows. The 26 columns are as follow: book_id, authors, average_rating, goodreads_book_id, country_code, description, format, image_url, is_ebook, isbn, isbn13, language_code, link, num_pages, publication_year, publisher, ratings_count, series, similar_books, title, title_without_series, URL and work_id.
Then, 3 datasets, ‘ratings.csv’, ‘books_tags.csv’, and ‘tags.csv’ are combined for collaborating as one rating dataset. The datasets are contributed by Philipp Spachtholz on Kaggle. The three data set contains following information:
ratings.csv (3 columns, 194,941 rows): book_id, user_id, rating
book_tags (3 columns and 999,912 rows): goodreads_book_id, tag_id, count
tags.csv(2 columns, 34,252 rows): Tag_id, tag_name
The combined rating data set has the same column name, “book_id” as the genre dataset, so the two datasets are now able to be used to build the book recommendation application.
The values of review rating can be biased because people tend to rate when they are really satisfied or when they are disappointed. The values of rating are the most important values to make a recommendation application since people get suggestions with the highly-rated books first. Before starting to create the algorithm, a distribution plot is driven to check if the dataset is biased with an average rating. Figure 1 shows that the data set is approximately normally distributed with an average of 3.90, so the values of average rating are not biased.
Figure 1. Distribution Plot of Average Rating
Then, a logistic regression model is used because the response variable, the genre is binary, the model checks if there is any relationship between the dependent variable, genre and independent variables, average rating, number of pages, publication year, rating count, text reviews count, and title of length. By using the glm function in R, a model is retrieved as following Figure 2.
Figure 2. Summary of Logistic Regression
Since the z-value for the number of pages and title length variable is very small, the coefficient of the two variables is significant. It explains the dataset well because the dependent variable is the two genres, Children’s book, and Fantasy book, the average number of pages of Children’s books are not supposed to be as much as adult books. Also, the average title length of Children’s books is 25% more than Fantasy books.
Next, the model is checked if the model meets the normality assumption by looking at the normal probability plot. Figure 3 shows the normal probability plot of residuals, and the normal distribution is shown.
Figure 3. Q-Q Plot of Residuals for Logistic Regression
This dataset will be used for an online recommendation system, so an average rating of books would be the most significant variable. A linear regression model is conducted on average rating versus other variables to check relationships between variables and the average rating. Since average rating is a numeric value, a multiple linear regression model is used and retrieved the following information.
Figure 4. Summary of Linear Regression
Since the p-value for the number of pages variable is very small, the coefficient of the variable is significant and concludes that number of pages has a linear relationship with the average rating. Then, checked if the model meets the normality assumption by looking at the normal probability plot in Figure 5. All of the points follow the straight line in the plot. This indicates that the residuals follow a normal distribution.
Figure 5. Q-Q Plot of Residuals for Linear Regression
Preprocessing of data
The data was processed in order to suit the machine learning algorithms better. This preprocessing included the following steps:
Binary encoding of categorical variables
Scaling of variables
Removal of missing values and columns with NA values
Figure 6. The collaborative Filtering Process
We chose collaborative filtering because it is widely used for recommendations. It assumes that if a person /A/ has the same opinion as a person /B/ on an issue, A is more likely to have B’s opinion on a different issue than that of a randomly chosen person.
Figure 6 simply illustrates how CF works. First, you should make a matrix in a certain format. Columns represent items and rows represent users. Then put this matrix in CF algorithms and get recommendations. ( Source: Item-Based Collaborative Filtering Recommendation Algorithms http://files.grouplens.org/papers/www10_sarwar.pdf )
We followed 3 steps for the recommendation. We referenced the process from Kaggle’s book recommendation example.
Data processing and exploration
Finding user neighbors
The first step is data processing and exploration. In this stage, we removed the duplicate ratings and then removed users who rated fewer than 3 books. Then we selected a subset of users for fast calculations. Then we explored the cleaned data set. The figure below shows title lengths with 5 or 7 words have slightly higher ratings. After that, we looked at which books are top-rated books and popular books. Next, we made a Matrix for the CF algorithm.
Figure 7. Exploratory Data Analysis
Figure 8. Top 10 top-rated books and top 10 popular books
Figure 9. Matix for UBCF
Second step is finding user neighbors. In this stage, we found similar users by comparing common books they liked. In this case, we set our current user as 794 and then found users who gave ratings to the same books. Then we normalized the user’s ratings and sorted users according to the similarity. For this, the Pearson correlation was used. Figure 11 shows similarities between the current user and 30 random users. For this plot, qgraph was used.
Figure 10. Finding user neighbors process
Figure 11. Similarities between users
Finally, in step three, based on similar users, the algorithm can recommend books that best fit the target user.
Figure 12. Best recommendations for the user
Model interpretation and explanation
To select the most suitable techniques for our project, we implemented both supervised and unsupervised techniques to investigate how well they perform on the dataset.
The datasets we used contain different kinds of data, from categorical, interval, ratio, to text, with more categorical data than other types of data. Therefore, classification models would fit our goal better.
We implemented three types of classification models, including Fast and Frugal decision tree analysis, xgboost and generalized linear model. To measure the performance of each model, ROC curve was created. As we can see from the Figure 13 below, all three supervised models do not have a good performance on the task, with tiny differences. The result suggested that supervised learning might not be the best approach for our project.
Figure 13. Performance ROC Curve
The result from the Fast and Frugal decision tree (Figure 14 below) indicated that the publication year, page number, genre, and how many times a book was reviewed are the most important variables in deciding whether a book would be recommended to a user.
Figure 14. Result from the Fast and Frugal Decision Tree
On the other hand, glm and xgboost identified ratings counts, page numbers, text review counts, together with publication year as the top four important features in their models.
Figure 15. Linear Vs. Non-Linear: glm vs. xgb
Provided that the preliminary analysis from the supervised learning offered some insights of the dataset, these models are not fulfilling our goal to recommend the next few books to read for our target users.
Moreover, to run the model, we eliminated a few variables because of the constraints of the data types in certain models. This procedure promoted the efficiency of training and building the model, yet we might have lost potential information in the process as well.
The nature of our app was to filter, and recommend books to the users, one of the state-of-art practices in building recommendation filtering systems is to use collaborative filtering algorithms, which mostly consider, as a clustering algorithm. Additionally, unsupervised learning outperformed supervised learning in finding potential existing patterns with different types of data. Therefore, we implemented a few unsupervised learning techniques on our dataset to see how they perform.
The ‘clValid’ package in R was used to carry out clustering on the dataset. Intuitively, the following parameters are the most relevant to users when it comes to recommending the best books - Average Rating of books, Genre, Format and Length of books (in terms of number of pages). These features were incorporated in our analysis after preprocessing which included binary encoding of categorical variables, scaling of the data points, and removal of NA/missing values.
We concluded that K-means clustering would better suit our data set instead of hierarchical clustering due to two main reasons:
The size of our dataset was large and binary encoding of variables would increase the number of variables that the algorithm would have to deal with.
K-means allows more flexibility with the clusters as compared to a hierarchical clustering model. If a data point needs to be reassigned to another cluster, doing that would only be possible with a K-means clustering model.
Using the clValid package in R, a cluster plot using the ‘kmeans’ method was obtained for 11 clusters. We validated 11 as the optimum number of clusters using silhouette and stability validation methods, searching over a range of 2 to 11. The highest silhouette score was associated with 11 clusters. A silhouette plot was also derived using the fviz_nbclust() function as shown below. The plot clearly shows the best number of clusters as 11.
Figure 16. The Silhouette Method for optimum number of clusters
Figure 17. Cluster plot
As seen in the plot above, the clusters obtained are very close to and overlap each other. Dimensionality Reduction methods such as UMAP and PCA can help resolve this issue. The UMAP function was used on our dataset to reduce it to two dimensions and produce cleaner and tighter clusters. The final clusters obtained after applying the UMAP function are as shown below:
Figure 18. Kmeans clustering based on UMAP transformed data
The green points in the above plot seem like outliers but they are actually a cluster of 10 points, a deeper analysis of which revealed that they belonged to the “Audio CD” format and their lengths were defined in terms of number of pages varying from 0-10. This aspect seems a bit peculiar especially considering the format of the books as audiobooks should, intuitively, not have any ‘pages’ or ‘page numbers’ associated with them.
The ‘kmeans’ function in R can also be used to fit the data. It is less complex and does not offer the same functionality as clValid but, with a little analysis, it allows for more interpretability of the different characteristics within each cluster. Here, the kmeans function was used with 11 clusters, and the data was filtered to obtain data points belonging to cluster 2. The books in this cluster belonged to the ‘Children’s Books’ genre and ‘Board Book’ format.
Figure 19. K-means clustering using ‘kmeans’ function
Nevertheless, it would not be easy for our app to rank all the books in the same cluster, therefore, the accuracy of the model might be compromised. On top of that, both models used dimension reduction methods, which reduce the interpretability of the models to the audience.
There are a few algorithms that proved to be effective, and Riester et al. (2020) found how noise levels impact the performance of each algorithm. As they suggested, UBCF and IBCF outperformed MFA in a more noisy dataset environment. In our app, the algorithm code contributed by Kaggle has both UBCF and IBCF algorithms. However, it would be better if we could conduct heuristic evaluations on our users to know better their experience.
Figure 20. Performance Graph: UBCF vs IBCF vs MFA
TF-IDF Word Cloud
Data: Book Description
Capture some info
Small data set
Figure 21. Word cloud Example
Application and Conclusions
First, we came up with a simple prototype. The prototype consists of 3 parts. In the first column, people can choose genres and rate books they like. In the second column, they can filter recommendations. Finally, they can see the recommended books. Using the arrow button, users can see more recommendations. Word clouds are made based on the book description and it enables users to understand what the recommended book is about in a short time.
Figure 22. Basic Prototype
Based on the shiny R example on Kaggle, we developed two new features, genre selection filtering and word cloud for recommendation books. As we used a different data set that consisted of Children’s books and Fantasy books, we made recommendations based on the user's selection of genre. After the user gets a recommendation, they can read word clouds and can understand the unique features of the recommended books. A screenshot of the current version of the app is shown below. You can find the test version here: https://edaxplor.shinyapps.io/book_v4/
Figure 23. Demo Video
From the demo, we were able to see UBCF works well for the book recommendation. However, we think further improvements can be made to the prototype by improving trust and persuasion. These are points that can be improved in the future.
Furthermore, conducting user testing would be an effective method to evaluate our service.
We used a username plus bookshelf as a title to increase the feeling of personalized service.
We separated the user’s bookshelf into two parts. First shelf is the ‘Reading Now’ shelf and the second shelf is the ‘To read' shelf. By doing so the algorithm can understand the user’s current and future interests.
By showing other users’ book self, we wanted to give users the feeling that this website is a book reading community. We think we can further add curated book sections for augmented AI human collaboration.
Users can Like or Dislike the recommendation. Based on this feedback, the algorithm can reflect the user’s preference and then apply it for the future recommendation
Figure 24. Detailed Prototype for Future Development
Genre Dataset: https://sites.google.com/eng.ucsd.edu/ucsdbookgraph/home
Mengting Wan, Julian McAuley, "Item Recommendation on Monotonic Behavior Chains", in RecSys'18. [bibtex]
Mengting Wan, Rishabh Misra, Ndapa Nakashole, Julian McAuley, "Fine-Grained Spoiler Detection from Large-Scale Review Corpora", in ACL'19. [bibtex]
Rating Dataset and Recommendation Algorithms: https://www.kaggle.com/philippsp/book-recommender-collaborative-filtering-shiny/code
Riesterer, N., Brand, D., & Ragni, M. (2020). Uncovering the Data-Related Limits of Human Reasoning Research: An Analysis based on Recommender Systems. arXiv preprint arXiv:2003.05196.
Choosing an optimal number of clusters: https://towardsdatascience.com/10-tips-for-choosing-the-optimal-number-of-clusters-277e93d72d92
Working and benefits of UMAP: https://towardsdatascience.com/how-exactly-umap-works-13e3040e1668