Information Technology, School of Computing and Informatics, College of Engineering and Technology, Wachemo University, Hosaena, Ethiopia.
Sci Rep. 2023 Mar 6;13(1):3693. doi: 10.1038/s41598-023-30987-0.
In the fields of machine learning and artificial intelligence, recommendation systems (RS) or recommended engines are commonly used. In today's world, recommendation systems based on user preferences assist consumers in making the best decisions without depleting their cognitive resources. They can be applied to a variety of things, including search engines, travel, music, movies, literature, news, gadgets, and dining. A lot of people utilize RS on social media sites like Facebook, Twitter, and LinkedIn, and it has proven beneficial in corporate settings like those at Amazon, Netflix, Pandora, and Yahoo. There have been numerous proposals for recommender system variations. However, certain techniques result in unfairly recommended things due to biased data because there are no established connections between the items and consumers. In order to solve the challenges mentioned above for new users, we propose in this work to employ Content-based Filtering (CBF) and Collaborative Filtering (CF) with semantic relationships to capture the relationships as knowledge-based book recommendations to readers in a digital library. When proposing things, patterns are more discriminative than single phrases. To capture the similarity of the books that the new user had retrieved, the patterns were grouped in a semantically equivalent manner using the Clustering method. The effectiveness of the suggested model is examined through a series of extensive tests employing Information Retrieval (IR) evaluation criteria. Recall Precision and F-Measure, two of the three widely used performance measuring metrics, were employed. The findings demonstrate that the suggested model performs noticeably better than cutting-edge models.
在机器学习和人工智能领域，推荐系统（RS）或推荐引擎被广泛应用。在当今世界，基于用户偏好的推荐系统可以帮助消费者在不消耗认知资源的情况下做出最佳决策。它们可以应用于各种事物，包括搜索引擎、旅行、音乐、电影、文学、新闻、小工具和餐饮。许多人在 Facebook、Twitter 和 LinkedIn 等社交媒体网站上使用 RS，它在亚马逊、网飞、潘多拉和雅虎等公司环境中也被证明是有益的。推荐系统有很多变体的提案。然而，由于数据存在偏差，某些技术会导致不公平的推荐，因为物品和消费者之间没有建立起固定的联系。为了解决上述新用户面临的挑战，我们在这项工作中提出，在数字图书馆中，使用基于内容的过滤（CBF）和基于语义关系的协同过滤（CF），将知识作为图书推荐的基础，为读者捕捉关系。在推荐物品时，模式比单一短语更具辨别力。为了捕捉新用户检索到的书籍之间的相似性，使用聚类方法以语义等效的方式对模式进行分组。通过使用信息检索（IR）评估标准进行的一系列广泛测试来检查所提出模型的有效性。使用了三个广泛使用的性能衡量指标中的两个，即召回精度和 F 度量。研究结果表明，所提出的模型明显优于最先进的模型。
