Book recommendation based on collaborative filtering technique / Norlina Mohd Sabri and Nurul Azeymasnita Jaffar
The book recommendation has been popularly adopted, especially for bookstore websites to improve their books suggestions to customers. The book recommendation system helps people who do not have enough personal experience to assess the alternatives offered by the website. As for the library, the com...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Book |
| Published: |
Universiti Teknologi MARA Cawangan Pulau Pinang,
2022-03.
|
| Subjects: | |
| Online Access: | Link Metadata |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The book recommendation has been popularly adopted, especially for bookstore websites to improve their books suggestions to customers. The book recommendation system helps people who do not have enough personal experience to assess the alternatives offered by the website. As for the library, the common Online Public Access Catalogue (OPAC) search could not acquire the spersonalisation of data as it is based on the basic search query. Thus, sometimes the users could not obtain the books according to their needs and interests. On the contrary, the book recommendation could produce results that are more spersonalised to the needs of users and could also cultivate a user's reading habit. Tstudy aims to investigate the efficiency of the book recommendation system based on the collaborative filtering technique. Collaborative filtering is one of the most adapted and powerful techniques for the recommendation system. The dataset used in this initial study was obtained from the Kaggle website and had been tested based on the 10-fold cross-validation technique. 1000 data had been chosen randomly, where 90% of the data were specified for the training phase, and another 10% were for the testing phase. The evaluation of the book recommendation prototype was based on the Precision, Recall and F-measure. In this initial research, the book recommender has successfully recommended the books with an acceptable performance of 80.38% F-measure value. However, the accuracy and efficiency of the recommender might further be increased if a larger volume of data is tested. Future works are to test a larger volume of data and investigate other well-known techniques in orderFuture works are to test a larger volume of data and investigate other well-known techniquesto identify the most efficient book recommender, especially for libraries. |
|---|---|
| Item Description: | https://ir.uitm.edu.my/id/eprint/62595/1/62595.pdf |