Maritzol Tenemaza (2022). User models for recommendation systems. In: Human Factors and Systems Interaction. AHFE Open Access, vol. 52, pp. 450–456. https://doi.org/10.54941/ahfe1002188
Currently, a user model or a representation of the user is used to recommend, personalize, predict, and manage inference in different recommender systems. Now, all the necessary user information is scattered. This document presents a systematic study to identify all the user information that recommendation systems use, represented or not by means of a model. This searching is important because it identifies the different data sources where user models are analyzed, and attributes and forms of user interaction are identified. The results presented are very useful for structuring a generalized user model.
Systematic review of the literature means identifying, evaluating, and inter- preting all the relevant research available for a particular research question, topic area, or phenomenon of interest (Keele, 2007). A systematic litera- ture review system is summarized in three main phases: planning the review, conducting the review, and reporting the review (Campoverde-Molina et al., 2020).
SYSTEMATIC REVIEW PROCESS The main research question is: what are the characteristics and structure for the user model?
Sources Selection The search is carried out in four databases of research publications: a) SCO- PUS, Scopus is the largest database of citations and abstracts of peer-reviewed literature. It offers a comprehensive overview of global research output. b) ACM Digital Library ACM (DL) is the world’s most comprehensive Library of full-text articles and bibliographic literature covering computing and information technology. c) IEEE Xplore It is a digital research library for the discovery and access to journal articles, conference proceedings, technical standards and related materials in computer science, electrical and electro- nic engineering, and related fields. d) Web of Science (WOS) platform with bibliographic information databases and information analysis resources that
allow evaluation and analysis of research performance. Its content is mul- tidisciplinary and provides information of a high academic and scientific level.
Parameter Definition The search parameters represent:
Definition of used terms.
Chain1: (User model OR user modeling OR user modelling)
Chain2: (User model OR User modeling) AND (Recommender system OR Recommendation system)
Place in the document where the terms are searched: To find the most items. Terms are searched for anywhere in the document.
The years of publication: The period from the year 2010 to the year 2020 has been restricted.
Types of articles: Articles in conferences and journals are considered. Books, technical reports, or any other type of publication are not considered.
Language: English strings have been specified to specifically search for documents in this language.
Relevance Determination For the presentation of results, the articles in each source are summarized for each year. If the publications are included or indexed in more than one source, duplications will be eliminated, therefore, the number of articles reviewed per year may be less than the number of articles reported. The content analysis is carried out manually to observe the required topics and their contribution to the context. Articles not important to the object of the investigation are discarded.
SYSTEMATIC REVIEW RESULTS Applying the steps established for the systematic review process, the follo- wing results are obtained. Table 1 shows the articles found with the search string 1, and Table 2 shows the articles found with the search string 2. Arti- cles are included that describe the different ways of generating user models, either considering the description of their attributes or techniques that are important for the definition of the model. The articles selected for each year are found in the Table 3.
FOUND TRENDS For Bakalov (Bakalov et al., 2013) the user model is represented by interests that are represented in a concept model, stores the exact value of the degree of interest in a scientific topic. Inzunza (Inzunza and Licea, 2018), Musto (Musto et al., 2018) Biamino (Biamino and Cena, 2011) Wang (Wang et al., 2010), (JIn et al., 2017) Coleho (Coleho et al., 2010) apply, Demographic characteristics: personal information such as name, nickname, height, weight.
452 Tenemaza
Table 1. Number of items found when applying string 1.
Año 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Total
5 4 2 2 7 5 7 7 8 4 1 52
Number of articles
Physical states: o physiological data of the person, such as heart assessment, sleep quality, keywords, stereotypes, and other information requested from the user etc., disabilities. Psychological aspects, such as the user’s personality or empathy. Interests and preferences. Affects: mood, emotions. Knowledge and objectives: What the user know and the objectives. User Connections and Relationships. Contacts (address, social network), Context information, Computing (software, hardware, networks), resources, location (physical - coordinates and address- and digital), time (season, time of day, type of day of the week), physical condition (weather, noise, brightness), and social relation- ships. Activity Information and Item Information: (Video, book, audio etc.). Dong (Dong et al., 2017), compare demographic user profiles, determine that young people are more active than older people. Normally people interact with people and similar age. That is, they analyze and compare demographic profiles of users on social networks.
Bogina (Bogina, 2017), studies temporal aspects of the user model. The goal is for the user model to continually adapt to new information available from the user. It proposes to detect and analyze the change of your interests, eliminate the included noise and memory usage.
Cena (Cena et al., 2016), performs an analysis of the data sources for the user model that contains data from the daily life of a person, where the view of the user’s context becomes relevant. They raise the user model not only from digital data, but from the use of ubiquitous technologies. Among the sensors proposed for the observation of user behavior are GPS, sensors,
User models for Recommendation Systems 453
smart vision, smartphones, smart devices. Ubiquitous technology will help to observe user behavior and actions in the real world, such as movements, tasks, recurring and repetitive habits. They could help determine sleeping patterns, eating patterns, levels of sedentary lifestyle, activity levels, use of media, etc. Likewise, the physical and mental states of the user, obtain them by tracking the user. From mobile devices, for example, blood pressure, tem- perature, glucose level, heart rate, etc. as well as persistent characteristics such as chronic diseases, relevant problems, hearing, movement, etc. The social relationships that the user has daily, obtained from the so-called daily, weekly meetings. The psychological record can be recorded from biosensors embed- ded in smart devices. The demographic characteristics of the user will be obtained by analyzing purchase invoices. Physical characteristics of the envi- ronment such as noise, congestion, climate, temperature; all this information is obtained from sensors.
Janowski (Janowski and Pinson, 2014), proposes a user model for rating behavior based on subject bias and error. Differences between subjects are important when analyzing and comparing people because bias can alter the results. The bias acts as a random variable within the ratings (For example, a bias is given, if one person is homophobic, it may turn out that everyone is homophobic) Lakiotaki and Tran (Lakiotaki et al., 2011, Tran et al., 2019), Saruk- kai (Sarukkai, 2013) Sang (Sang et al., 2015), Yang (Yan et al., 2016), Piao (Piao and Breslin, 2018), Ha (Ha et al., 2015), Musto (Musto et al., 2018), Sang (Sang et al., 2015). Yang (Yang et al., 2014), Ha (Ha et al., 2015), Sarukkai (Sarukkai, 2013), Musto (Musto et al., 2018) Piao (Piao and Breslin, 2018), Lai (Lai et al., 2018) considers the social networks Twitter, Facebook, and LinkedIn, YouTube Twitter, Google+. They analyze inten- tions and preferences influenced by intrinsic interests from user behaviors in social networks considering rankings. Alnahhas (Alnahhas and Alkhatib, 2020), Barros (Barros et al., 2012, Lai et al., 2018), analyze emotions, wri- ting styles and tendencies, variation in texts and variation of emotions on written texts, they propose to build a user model based on the conceptual analysis of texts, to include contextual information of concepts to identify the user’s interests: the analysis include common sense in words and phrases analyzing semantic relationships in the knowledge. Guo (Guo et al., 2016), show a computational model that considers social influence in recommen- der systems. People, directly or indirectly, are influenced by the feelings and actions of others. Social factors such as personality, expertise, interpersonal relationships, and similar preferences amplify the effect of social influence. Wu (Wu, 2012), user preference is driven by two now dynamic relationsh- ips 1) user-item relationship, and 2) social network structure reflected by user-user behavior. They generate a user model, considering the historical behaviors in the two types of relationships. They consider hemophilia (love for equals), closeness and social influence.
Madureira (Madureira et al., 2014), to know the decisions and restrictions of the user, from user actions on the system. Sheng (Sheng and Liu, 2010), wants to know the needs of the user, basing his work on the user’s search history. Abdar (Abdar and Yen, 2017), identifies the user’s context, relating
454 Tenemaza
internal and external factors of the user, internal are considered: priority, preference, attitude and alternatives and external are considered searches, messages, GPS, sensors, city data and date.
Similarly, Paolino (Paolino et al., 2010), Takama (Takama et al., 2018), Kompan (Kompan et al., 2019), and Berger (Berger and Kompan, 2019) base their analysis on stable user characteristics or favorite items called long-term and the latest content delivered called short term or user actions. short-term user. The most common are the short term to model context and behavior. Kompan (Kompan et al., 2019) and Berger (Berger and Kompan, 2019) discuss session abandonment. Takama (Takama et al., 2018) recommends less popular but important items for a small number of people. Guntuku (Guntuku et al., 2016), their goal is to identify a) personality recognition and b) personality perception. based on the inference from the likes of images and/or request a self-assessment from the user.
People with special needs are also considered in the exhibition of user models to represent them. Thus, Kurschl (Kurschl et al., 2014), expose a user model for people with special needs. They look for patterns of interaction based on touch and without touch. They present a user model (stores infor- mation about the user experience and interaction concepts), an input model (contains all the information about the input methods and their configura- tion). They describe the way to control the application, an adaptation model (describes how the content can be adapted to optimize the presentation to the user), and a domain model (describes the content and organization of the domain). They propose the user interface with special needs. Also, Arbelaitz (Arbelaitz et al., 2016), create a social network for people with cognitive disabilities.
CONCLUSION
As it has been observed, there is a lot of research on attributes that are necessary to represent the user, the different sources and the possible characte- ristics that represent the user are described. As you can see the information is scattered, many researchers focus their attention on specific aspects without looking at a specific model. This opens the way to gather all the infor- mation generated and observe gaps and in a next step represent a general model that represents the user, which will mean a contribution to current recommendation systems.
REFERENCES
Abdar, M. & Yen, N. Y. 2017. Design of a universal user model for dynamic crowd preference sensing and decision-making behavior analysis. IEEE Access, 5, 24842–24852. Alnahhas, A. & Alkhatib, B. 2020. Augmented context-based conceptual user
modeling for personalized recommendation system in online social networks. International Journal of Cognitive Informatics and Natural Intelligence (IJCINI), 14, 1–19.
User models for Recommendation Systems 455
Arbelaitz, O., MartÍnez-Otzeta, J. M. & Muguerza, J. 2016. User modeling in a social network for cognitively disabled people. Journal of the Association for Information Science and Technology, 67, 305–317. Bakalov, F., Meurs, M.-J., König-Ries, B., Sateli, B., Witte, R., Butler, G. & Tsang, A.
An approach to controlling user models and personalization effects in recommen- der systems. Proceedings of the 2013 international conference on Intelligent user interfaces, 2013. 49–56. Barros, L., RodrÍguez, P. & Ortigosa, A. Emotion recognition in texts for user model augmenting. Proceedings of the 13th International Conference on Interacción Persona-Ordenador, 2012. 1–6. Berger, P. & Kompan, M. 2019. User modeling for churn prediction in E-commerce.
IEEE Intelligent Systems, 34, 44–52. Biamino, G. & Cena, F. Social awareness and user modeling to improve objects intelligence. 2011 IEEE/WIC/ACM International Conferences on Web Intelligence and Intelligent Agent Technology, 2011. IEEE, 118–121. Bogina, V. Addressing Temporal Aspects in User Modelling. Adjunct Publication of
the 25th Conference on User Modeling, Adaptation and Personalization, 2017. 113–114. Campoverde-Molina, M., Luján-Mora, S. & García, L. V. 2020. Empirical studies on web accessibility of educational websites: A systematic literature review. IEEE Access, 8, 91676–91700. Cena, F., Gena, C. & Picardi, C. An Experimental Study in Cross-Representation
Mediation of User Models. Proceedings of the 2016 Conference on User Modeling Adaptation and Personalization, 2016. 289-290. Coleho, B., Martins, C. & Almeida, A. Web intelligence in tourism: User modeling and recommender system. 2010 IEEE/WIC/ACM International Conference on Web Intelligence and Intelligent Agent Technology, 2010. IEEE, 619–622. Dong, Y., Chawla, N. V., Tang, J., Yang, Y. & Yang, Y. 2017. User modeling
on demographic attributes in big mobile social networks. ACM Transactions on Information Systems (TOIS), 35, 1–33. Guntuku, S. C., Zhou, J. T., Roy, S., Lin, W. & Tsang, I. W. 2016. Who likes what and, why? ‘insights into modeling users’ personality based on image ‘likes. IEEE Transactions on Affective Computing, 9, 130–143. Guo, J., Zhu, Y., Li, A., Wang, Q. & Han, W. 2016. A social influence approach for
group user modeling in group recommendation systems. IEEE Intelligent systems, 31, 40–48. HA, I., OH, K.-J. & JO, G.-S. 2015. Personalized advertisement system using social relationship based user modeling. Multimedia Tools and Applications, 74, 8801–8819. Inzunza, S. & Licea, G. 2018. GUMCARS: General User Model for Context-Aware
Recommender Systems. Computing and Informatics, 37, 1149–1183. Janowski, L. & Pinson, M. Subject bias: Introducing a theoretical user model. 2014 Sixth International Workshop on Quality of Multimedia Experience (QoMEX), 2014. IEEE, 251–256. Jin, X., Li, H., Liu, Y., Wen, L., Fan, Y. & Hao, Q. Research on User Model Constru-
ction for Individualized Knowledge Recommendation. 2017 13th International Conference on Semantics, Knowledge and Grids (SKG), 2017. IEEE, 165–169. Keele, S. 2007. Guidelines for performing systematic literature reviews in software engineering. Citeseer. Kompan, M., Kassak, O. & Bielikova, M. 2019. The Short-term User Modeling for
Predictive Applications. Journal on Data Semantics, 8, 21–37.
456 Tenemaza
Kurschl, W., Augstein, M., Burger, T. & Pointner, C. 2014. User modeling for people with special needs. International Journal of Pervasive Computing and Communications. Lai, K.-H., Yen, N. Y. & Hung, J. C. User modeling based on sentiment analysis.
2018 9th International conference on information technology in medicine and education (ITME), 2018. IEEE, 1072–1076. Lakiotaki, K., Matsatsinis, N. F. & Tsoukias, A. 2011. Multicriteria user modeling in recommender systems. IEEE Intelligent Systems, 26, 64–76. Madureira, A., Cunha, B., Pereira, J. P., Pereira, I. & Gomes, S. An architecture for
user modeling on Intelligent and Adaptive Scheduling Systems. 2014 Sixth World Congress on Nature and Biologically Inspired Computing (NaBIC 2014), 2014. IEEE, 103–108. Musto, C., Semeraro, G., Lovascio, C., De Gemmis, M. & Lops, P. A framework for holistic user modeling merging heterogeneous digital footprints. Adjunct Publica- tion of the 26th Conference on User Modeling, Adaptation and Personalization, 2018. 97–101. Paolino, L., Sebillo, M., Tortora, G., Vitiello, G., Martellone, A. M. & Tacconi,
D. 2010. Dynamic User Modeling for Personalized Advertisement Delivery on Mobile Devices. Management of the Interconnected World. Springer. Piao, G. & Breslin, J. G. 2018. Inferring user interests in microblogging social networks: a survey. User Modeling and User-Adapted Interaction, 28, 277–329. Sang, J., Deng, Z., Lu, D. & Xu, C. 2015. Cross-osn user modeling by homogene-
ous behavior quantification and local social regularization. IEEE Transactions on Multimedia, 17, 2259–2270. Sarukkai, R. R. Real-time user modeling and prediction: examples from Youtube.
Proceedings of the 22nd International Conference on World Wide Web, 2013. 775–776. Sheng, J. & Liu, S. 2010. A Knowledge Recommend System Based on User Model.
International Journal of Digital Content Technology and its Applications, 4, 168–173. Takama, Y., Chen, Y.-S., Misawa, R. & Ishikawa, H. 2018. Analyzing potential of personal values-based user modeling for long tail item recommendation. Journal of Advanced Computational Intelligence and Intelligent Informatics, 22, 506–513. Tran, C., Kim, J.-Y., Shin, W.-Y. & Kim, S.-W. 2019. Clustering-based colla-
borative filtering using an incentivized/penalized user model. IEEE Access, 7, 62115–62125. Wang, Q., Yuan, X. & Sun, M. Collaborative filtering recommendation algorithm based on hybrid user model. 2010 Seventh International Conference on Fuzzy Systems and Knowledge Discovery, 2010. IEEE, 1985–1990. Wu, J. 2012. Advances in K-means clustering: a data mining thinking, Springer
Science & Business Media. Yan, M., SANG, J., XU, C. & HOSSAIN, M. S. 2016. A unified video recommendation by cross-network user modeling. ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM), 12, 1–24. Yang, X., Guo, Y., Liu, Y. & Steck, H. 2014. A survey of collaborative filtering based
social recommender systems. Computer communications, 41, 1–10.
LISTED IN THE AHFE DIGITAL BOOKSTORE CATALOGUE
Explore the complete digital volume
Human Factors and Systems Interaction
Read this individual research paper free through the original publisher, or explore the full digital proceedings book.
ISBN 978-1-958651-28-5
Explore this ISBN in the AHFE Bookstore ↗Listed in the July 2026 bookstore catalogue; search by ISBN to confirm current availability · Digital book purchase optional