Editing Recommender system (section)

==Technologies==

=== Session-based recommender systems ===
These recommender systems use the interactions of a user within a session<ref name="a">{{cite arXiv|last1=Hidasi|first1=Balázs|last2=Karatzoglou|first2=Alexandros|last3=Baltrunas|first3=Linas|last4=Tikk|first4=Domonkos|date=2016-03-29|title=Session-based Recommendations with Recurrent Neural Networks|class=cs.LG|eprint=1511.06939}}</ref> to generate recommendations. Session-based recommender systems are used at YouTube<ref name="yt">{{cite arXiv|last1=Chen|first1=Minmin|last2=Beutel|first2=Alex|last3=Covington|first3=Paul|last4=Jain|first4=Sagar|last5=Belletti|first5=Francois|last6=Chi|first6=Ed|title=Top-K Off-Policy Correction for a REINFORCE Recommender System|year=2018|class=cs.LG|eprint=1812.02353}}</ref> and Amazon.<ref name="amzn">{{Cite book|last1=Yifei|first1=Ma|last2=Narayanaswamy|first2=Balakrishnan|last3=Haibin|first3=Lin|last4=Hao|first4=Ding|title=Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining |chapter=Temporal-Contextual Recommendation in Real-Time |year=2020|pages=2291–2299|publisher=Association for Computing Machinery|doi=10.1145/3394486.3403278|isbn=978-1-4503-7998-4|s2cid=221191348|doi-access=free}}</ref> These are particularly useful when history (such as past clicks, purchases) of a user is not available or not relevant in the current user session. Domains where session-based recommendations are particularly relevant include video, e-commerce, travel, music and more. Most instances of session-based recommender systems rely on the sequence of recent interactions within a session without requiring any additional details (historical, demographic) of the user. Techniques for session-based recommendations are mainly based on generative sequential models such as [[Recurrent neural network|recurrent neural networks]],<ref name="a" /><ref>{{Cite book|last1=Hidasi|first1=Balázs|last2=Karatzoglou|first2=Alexandros|title=Proceedings of the 27th ACM International Conference on Information and Knowledge Management |chapter=Recurrent Neural Networks with Top-k Gains for Session-based Recommendations |date=2018-10-17|chapter-url=https://doi.org/10.1145/3269206.3271761|series=CIKM '18|location=Torino, Italy|publisher=Association for Computing Machinery|pages=843–852|doi=10.1145/3269206.3271761|arxiv=1706.03847|isbn=978-1-4503-6014-2|s2cid=1159769}}</ref> [[Transformer (deep learning architecture)|transformers]],<ref>{{cite arXiv|last1=Kang|first1=Wang-Cheng|last2=McAuley|first2=Julian|title=Self-Attentive Sequential Recommendation|year=2018|class=cs.IR|eprint=1808.09781}}</ref> and other deep-learning-based approaches.<ref>{{Cite book|last1=Li|first1=Jing|last2=Ren|first2=Pengjie|last3=Chen|first3=Zhumin|last4=Ren|first4=Zhaochun|last5=Lian|first5=Tao|last6=Ma|first6=Jun|title=Proceedings of the 2017 ACM on Conference on Information and Knowledge Management |chapter=Neural Attentive Session-based Recommendation |date=2017-11-06|chapter-url=https://doi.org/10.1145/3132847.3132926|series=CIKM '17|location=Singapore, Singapore|publisher=Association for Computing Machinery|pages=1419–1428|doi=10.1145/3132847.3132926|arxiv=1711.04725|isbn=978-1-4503-4918-5|s2cid=21066930}}</ref><ref>{{Cite book|last1=Liu|first1=Qiao|last2=Zeng|first2=Yifu|last3=Mokhosi|first3=Refuoe|last4=Zhang|first4=Haibin|title=Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining |chapter=STAMP |date=2018-07-19|chapter-url=https://doi.org/10.1145/3219819.3219950|series=KDD '18|location=London, United Kingdom|publisher=Association for Computing Machinery|pages=1831–1839|doi=10.1145/3219819.3219950|isbn=978-1-4503-5552-0|s2cid=50775765}}</ref>

=== Reinforcement learning for recommender systems ===

The recommendation problem can be seen as a special instance of a reinforcement learning problem whereby the user is the environment upon which the agent, the recommendation system acts upon in order to receive a reward, for instance, a click or engagement by the user.<ref name="yt" /><ref name="srl">{{cite arXiv|last1=Xin|first1=Xin|last2=Karatzoglou|first2=Alexandros|last3=Arapakis|first3=Ioannis|last4=Jose|first4=Joemon|title=Self-Supervised Reinforcement Learning for Recommender Systems|year=2020|class=cs.LG|eprint=2006.05779}}</ref><ref name="sQ">{{Cite journal|last1=Ie|first1=Eugene|last2=Jain|first2=Vihan|last3=Narvekar|first3=Sanmit|last4=Agarwal|first4=Ritesh|last5=Wu|first5=Rui|last6=Cheng|first6=Heng-Tze|last7=Chandra|first7=Tushar|last8=Boutilier|first8=Craig|title=SlateQ: A Tractable Decomposition for Reinforcement Learning with Recommendation Sets|url=https://research.google/pubs/pub48200/|journal=Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence (IJCAI-19)|year=2019|pages=2592–2599}}</ref> One aspect of reinforcement learning that is of particular use in the area of recommender systems is the fact that the models or policies can be learned by providing a reward to the recommendation agent. This is in contrast to traditional learning techniques which rely on supervised learning approaches that are less flexible, reinforcement learning recommendation techniques allow to potentially train models that can be optimized directly on metrics of engagement, and user interest.<ref name="jd">{{Cite book|last1=Zou|first1=Lixin|last2=Xia|first2=Long|last3=Ding|first3=Zhuoye|last4=Song|first4=Jiaxing|last5=Liu|first5=Weidong|last6=Yin|first6=Dawei|title=Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining |chapter=Reinforcement Learning to Optimize Long-term User Engagement in Recommender Systems |chapter-url=https://dl.acm.org/doi/10.1145/3292500.3330668|series=KDD '19|year=2019|pages=2810–2818|doi=10.1145/3292500.3330668|arxiv=1902.05570|isbn=978-1-4503-6201-6|s2cid=62903207}}</ref>

=== Multi-criteria recommender systems ===
Multi-criteria recommender systems (MCRS) can be defined as recommender systems that incorporate preference information upon multiple criteria. Instead of developing recommendation techniques based on a single criterion value, the overall preference of user u for the item i, these systems try to predict a rating for unexplored items of u by exploiting preference information on multiple criteria that affect this overall preference value. Several researchers approach MCRS as a multi-criteria decision making (MCDM) problem, and apply MCDM methods and techniques to implement MCRS systems.<ref>{{Cite journal |last1 = Lakiotaki |first1 = K. |last2 = Matsatsinis |last3 = Tsoukias |first3 = A |title = Multicriteria User Modeling in Recommender Systems |journal = IEEE Intelligent Systems |volume = 26 |issue = 2 |pages = 64–76 |doi=10.1109/mis.2011.33|date = March 2011 |citeseerx = 10.1.1.476.6726 |s2cid = 16752808 }}</ref> See this chapter<ref>{{cite web |url=http://ids.csom.umn.edu/faculty/gedas/NSFCareer/MCRS-chapter-2010.pdf |title=Multi-Criteria Recommender Systems |author=Gediminas Adomavicius |author2=Nikos Manouselis |author3=YoungOk Kwon  |archive-url=https://web.archive.org/web/20140630021251/http://ids.csom.umn.edu/faculty/gedas/NSFCareer/MCRS-chapter-2010.pdf |archive-date=2014-06-30 }}</ref> for an extended introduction.

=== Risk-aware recommender systems ===
The majority of existing approaches to recommender systems focus on recommending the most relevant content to users using contextual information, yet do not take into account the risk of disturbing the user with unwanted notifications. It is important to consider the risk of upsetting the user by pushing recommendations in certain circumstances, for instance, during a professional meeting, early morning, or late at night. Therefore, the performance of the recommender system depends in part on the degree to which it has incorporated the risk into the recommendation process. One option to manage this issue is ''DRARS'', a system which models the context-aware recommendation as a [[Multi-armed bandit|bandit problem]]. This system combines a content-based technique and a contextual bandit algorithm.<ref name="Bouneffouf2013">{{cite thesis | type=Ph.D. thesis | last = Bouneffouf  | first = Djallel | title= DRARS, A Dynamic Risk-Aware Recommender System | publisher = Institut National des Télécommunications | year = 2013 | url = http://tel.archives-ouvertes.fr/tel-01026136/fr/}}</ref>

=== Mobile recommender systems ===
{{further|Location based recommendation}}
Mobile recommender systems make use of internet-accessing [[Smartphone|smartphones]] to offer personalized, context-sensitive recommendations. This is a particularly difficult area of research as mobile data is more complex than data that recommender systems often have to deal with. It is heterogeneous, noisy, requires spatial and temporal auto-correlation, and has validation and generality problems.<ref name="taxirecommender">{{cite conference |author1=Yong Ge |author2=Hui Xiong |author3=Alexander Tuzhilin |author4=Keli Xiao |author5=Marco Gruteser |author6=Michael J. Pazzani |year = 2010 |title = An Energy-Efficient Mobile Recommender System |conference = Proceedings of the 16th ACM SIGKDD Int'l Conf. on Knowledge Discovery and Data Mining |url = http://www.winlab.rutgers.edu/~gruteser/papers/KDD10.pdf |publisher = [[Association for Computing Machinery|ACM]] |location = [[New York City|New York City, New York]] |pages = 899–908 |access-date = 2011-11-17 }}
</ref>

There are three factors that could affect the mobile recommender systems and the accuracy of prediction results: the context, the recommendation method and privacy.<ref>{{cite journal |last1=Pimenidis |first1=Elias |last2=Polatidis |first2=Nikolaos |last3=Mouratidis |first3=Haralambos |title=Mobile recommender systems: Identifying the major concepts |journal=Journal of Information Science |volume=45 |issue=3 |date=3 August 2018 |pages=387–397 |doi=10.1177/0165551518792213|arxiv=1805.02276 |s2cid=19209845 }}</ref> Additionally, mobile recommender systems suffer from a transplantation problem – recommendations may not apply in all regions (for instance, it would be unwise to recommend a recipe in an area where all of the ingredients may not be available).

One example of a mobile recommender system are the approaches taken by companies such as [[Uber]] and [[Lyft]] to generate driving routes for taxi drivers in a city.<ref name="taxirecommender"/> This system uses GPS data of the routes that taxi drivers take while working, which includes location (latitude and longitude), time stamps, and operational status (with or without passengers). It uses this data to recommend a list of pickup points along a route, with the goal of optimizing occupancy times and profits.

=== Generative recommenders ===
Generative recommenders (GR) represent an approach that transforms recommendation tasks into [[Seq2seq|sequential transduction]] problems, where user actions are treated like tokens in a generative modeling framework. In one method, known as HSTU (Hierarchical Sequential Transduction Units),<ref>{{cite arXiv |last1=Zhai |first1=Jiaqi |title=Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Generative Recommendations |date=2024-05-06 |eprint=2402.17152 |last2=Liao |first2=Lucy |last3=Liu |first3=Xing |last4=Wang |first4=Yueming |last5=Li |first5=Rui |last6=Cao |first6=Xuan |last7=Gao |first7=Leon |last8=Gong |first8=Zhaojie |last9=Gu |first9=Fangda|class=cs.LG }}</ref> high-[[cardinality]], non-stationary, and streaming datasets are efficiently processed as sequences, enabling the model to learn from trillions of parameters and to handle user action histories orders of magnitude longer than before. By turning all of the system’s varied data into a single stream of tokens and using a custom [[Attention (machine learning)|self-attention]] approach instead of [[Neural network (machine learning)|traditional neural network layers]], generative recommenders make the model much simpler and less memory-hungry. As a result, it can improve recommendation quality in test simulations and in real-world tests, while being faster than previous [[Transformer (deep learning architecture)|Transformer]]-based systems when handling long lists of user actions. Ultimately, this approach allows the model’s performance to grow steadily as more computing power is used, laying a foundation for efficient and scalable “[[Foundation model|foundation models]]” for recommendations.