Zum Inhalt springen
Technische Universität München

Seminar in Winter Semester 2022/23:

Topics in Recommender Systems

(Ashmi Banerjee)

News

[19.01.2023] Presentation schedule announced! They will be held online over MS Teams! You should have received email with details.

[09.10.2022] Topics assigned to participants, see below.

[29.06.2022] Web page online. This seminar will be based on the new 3rd Edition of the Recommender Systems Handbook. You should be able to check out the electronic version of the handbook from TUM networks or by logging in with your TUM account.

Information

  • This seminar is for students in the Informatics Master program (module IN2107)
    • The seminar will be conducted in English language
    • Prerequisites are a Bachelor's degree in computer science or related field
  • Students are expected to write a paper and give a presentation about the given topic area
    • Just summarizing related work is only the foundation of your paper, you need to show an own contribution beyond the summarization of other work. This contribution can be a new classification scheme, assessment and/or comparison of existing work, coming up with some novel conceptual ideas, design of an algorithm idea, mock-up of a new user interface, sketching new application areas and/or similar contributions
    • Your paper can be very focussed, concentrating on a (small) subset of the given area. You can for example first give an brief overview of the topic area and then dig deeper on a selected aspect
    • You need to search for suitable literature in addition to the stated references. Orient yourself to the given references or other research papers for structure/contents of your own paper. You need to cite the literature your work is based on and clearly indicate when you are adopting or paraphrasing other work
    • To find additional literature, Google Scholar (scholar.google.com) is a valuable source. You can not only find literature based on keywords, but also find papers that have cited a certain paper. To do so, follow the "cited by ..." links of your most relevant articles.
  • Information for paper
    • 12-14 pages (including references) in English in the newer ACM format with a single column, and not the older one with two columns
    • State your name, affiliation and email address (as the only author), use your own keywords and include a short abstract
    • No postal address or telephone number, no permission block, copyright line or page numbering, no categories and subject descriptors or general terms
    • References and citations need to be in the correct format (but usage of a LaTex BIB file is optional)
    • Acknowledgments or appendix are optional and not expected
  • Information for presentation
    • Duration is 15-20 minutes, including questions&answers
    • The talk should be given freely, i.e. not completely read out from a script (in English)
    • You should present slides electronically in any format (e.g. Powerpoint or PDF)
    • You can use the TUM powerpoint template or your own format for the slides
  • All topics are advised by Ashmi Banerjee (ashmi.banerjee AT tum.de)
  • Grading will be based on both the paper and the presentation (approximately equal weight)
  • Prerequisites for credits:
    • Submit the paper in acceptable quality until the stated deadline
    • Give a presentation of acceptable quality on the assigned date
    • Attend all presentation meetings and participate in the discussion

Procedure

  • Information meeting on Thu, 20.10.2022, 16:00 (remote/room tbd) with focus on the paper (but you are free to start earlier working on your paper after the matching is complete and you have been assigned a place and topic!)
  • (Optional) Midterm/intermediate submission until 16.11.2022, 23:59.
    • Submit the intermediate draft (PDF)  until 16.11.2022 via email to Ashmi Banerjee (ashmi.banerjee AT tum.de) (sending a Dropbox link or something similar is also possible)
  • Submission of your paper in the correct format until Sunday, 15.01.2023, 23:59 (no extensions!)
    • Submit the PDF and also the source file(s) (TeX or Word) via Email to Ashmi Banerjee (ashmi.banerjee AT tum.de) (sending a Dropbox link or something similar is also possible)
  • The presentations will be held on 31.01., 01.02. and 02.02.2023 each starting at 16:00 (room tbd)

Registration

Presentation Schedule

Day 1: 31.01.2023 from 16:00 over MS Teams

  • Omar Labib [Cross-domain approaches to Recommender Systems]
  • Xiaofeng Deng [Natural Language techniques for Recommender Systems]
  • Evgeniya Sukhodolskaya [Human Decision Making in Interactive Recommender Systems]
  • Hassaan Wasim [Tourism Recommender Systems]

Day 2: 02.02.2023 from 16:00 over MS Teams

  • Viet Duc Mai [Touristic Region Detection]
  • Yuhang Tang [Adversarial Machine Learning for Recommender Systems]
  • Fitri Nur Aisyah [Evaluating the User Experience of Recommender Systems]
  • Furkan Çelik [Automatic Preference Elicitation from Social Media]

Topics and Literature

Foundation article for all topics:

Some more example resources for each topic have been added. Students are encouraged to use addtional relevant resources besides the ones mentioned for their work.

1. Adversarial Machine Learning for Recommender Systems (Yuhang Tang)

Recommender Systems are built with a focus to exploit user interaction data and side-information so that they can provide improved results. All these algorithms have the same underlying fundamental assumption of “data stationarity”, that is, both training and test data are sampled from a similar distribution. However, in an adversarial setting, an intelligent and adaptive attacker deliberately manipulates the data such that this stationarity assumption is violated, thus compromising the RS’s integrity. The integrity of the RS is of paramount significance as it is often used to make life-affecting decisions. Hence, it is important to understand how to build an efficient yet secure RS, protected from the clutches of an attacker.

  • Yashar Deldjoo, Tommaso Di Noia, and Felice Antonio Merra. 2020. Adversarial Machine Learning in Recommender Systems (AML-RecSys). Proceedings of the 13th International Conference on Web Search and Data Mining. Association for Computing Machinery, New York, NY, USA, 869–872.
    https://dl.acm.org/doi/abs/10.1145/3336191.3371877

2. People-to-People Reciprocal Recommender Systems 

While most recommender systems recommend a list of items, they rarely take into consideration the other side i.e. they are solely determined by the choices of the user seeking the recommendation. However, in case of social-networking platforms or dating platforms, the success of the system is determined if both the concerned parties are interested. This topic explores the different scenarios of a RS where reciprocation is expressed and discusses the distinctive nature of the reciprocal recommenders.

  • Iván Palomares, Carlos Porcel, Luiz Pizzato, Ido Guy, Enrique Herrera-Viedma, Reciprocal Recommender Systems: Analysis of state-of-art literature, challenges and opportunities towards social recommendation, Information Fusion, Volume 69, 2021, Pages 103-127, ISSN 1566-2535,
    https://doi.org/10.1016/j.inffus.2020.12.001

3. Natural Language techniques for Recommender Systems (Xiaofeng Deng)

Often RS receives enormous amounts of texts as input and it is important to analyze these texts using advanced Natural Language Processing Algorithms (NLP). This topic explores the variety of textual inputs that can be accepted by a RS and identify the cases where NLP can potentially assist in solving the problems.

  • Weiwei Guo, Huiji Gao, Jun Shi, Bo Long, Liang Zhang, Bee-Chung Chen, and Deepak Agarwal. 2019. Deep Natural Language Processing for Search and Recommender Systems. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD '19). Association for Computing Machinery, New York, NY, USA, 3199–3200.
    https://doi.org/10.1145/3292500.3332290

4. Fairness in Recommender Systems (Jaeyeop Chung)

 A common concern prevalent in the recent years is whether these Recommender Systems allocate resources fairly or they are biased, resulting in an unfair distribution of items or resources. RS often fall prey to Popularity Bias where certain results are shown more frequently to the users despite their poor quality. The topic studies the fairness in Recommender Systems, possible use-cases of biases and how they can be redesigned to mitigate them.

  • Jurek Leonhardt, Avishek Anand, and Megha Khosla. 2018. User Fairness in Recommender Systems. In Companion Proceedings of the The Web Conference 2018 (WWW '18). International World Wide Web Conferences Steering Committee, Republic and Canton of Geneva, CHE, 101–102.
    https://doi.org/10.1145/3184558.3186949

 

5. Cross-domain approaches to Recommender Systems (Omar Labib)

When stored user information across multiple domains is used to recommend items to the user, it is referred to as cross-domain approach to RS. Often websites like Alibaba or Amazon use prior user history while recommending items to the user. However, it becomes increasingly difficult for companies belonging to a single domain e.g. Spotify, Netflix etc. This topic explores the different cross-domain recommender systems, their challenges owing to knowledge transfers across multiple domains, and evaluates them.
 

  • Muhammad Murad Khan, Roliana Ibrahim, and Imran Ghani. 2017. Cross Domain Recommender Systems: A Systematic Literature Review. ACM Comput. Surv. 50, 3, Article 36 (May 2018), 34 pages.
    doi.org/10.1145/3073565

 

6. Group Recommendation Techniques (Phuong Anh Nguyen)

When recommending items to a group of users instead of a single user, the preferences of all group members have to be taken into account. Different preference aggregation strategies exist for this purpose. Basic approaches such as calculating an average preference of all users are easy to realize but might not be optimal. For example, if one user really dislikes a certain item, this item should not be recommended even if the majority of group members likes the item. It is important to say that there is no perfect way to aggregate the individual preferences. Instead, the group's intrinsic characteristics and the problem's nature have to be considered.

* Jameson and Smyth (2007): Recommendation to Groups

* Masthoff (2015): Group Recommender Systems: Aggregation, Satisfaction and Group Attributes

* Alvarado Rodriguez et al. (2022): A Systematic Review of Interaction Design Strategies for Group Recommendation systems

7. Multi-stakeholder recommender systems (Hassaan Wasim)

Often, recommender systems are optimized for recommendations to best suit the users' interests and preferences only. However, other stakeholders are involved in real work recommender applications. For example, in e-Commerce, sellers may want to maximize their profit, and in tourism recommendation, suggesting the most popular places may lead to overtourism. In addition, objectives such as coverage, diversity and fairness play an important role when generating recommendations. Therefore, recommender systems need to balance possible conflicting interests of different parties and take ethical considerations into account.

* Abdollahpouri et al. (2020): Multistakeholder Recommendation: Survey and Research Directions

* Zheng (2017): Multi-Stakeholder Recommendation: Applications and Challenges

* Milano et. al (2020): Recommender Systems and their Ethical Challenges

 

8. Human Decision Making in Interactive Recommender Systems (Evgeniya Sukhodolskaya)

The user interface influences how people make decision. For example, when searching for products it is not only important to list appropriate items but also consider how to present them. The sequence of items plays a role, adding images may increase sale, or adding an inferior item may actually persuade customers to buy a target product. This topic should discuss some aspects of the interplay between user interface design and human decision making in recommender systems.

* Jameson et al. (2015): Human Decision Making and Recommender Systems

* Bollen at al. (2010): Understanding Choice Overload in Recommender Systems

* Ekstrand and Willemsen (2016): Behaviorism is Not Enough: Better Recommendations Through Listening to Users

9. Evaluating the User Experience of Recommender Systems (Fitri Nur Aisyah)

Recommender systems recommend products and services such as movies or restaurants to users. The quality of a recommender system is not only determined by the accuracy of the recommendations. The overall user experience (UX) when interacting with the recommender system is a very important factor which decides if people like using the recommender system. According to Hassenzahl (2008), UX is "a momentary, primarily evaluative feeling (good-bad) while interacting with a product or service. Good UX is the consequence of fulfilling the human needs for autonomy, competence, stimulation (self-oriented) through interacting with the product or service (i.e. hedonic quality)." Many methods exist to evaluate the UX of recommender systems but also the usability of the system's user interfaces which is one very important aspect of UX.

* Knijnenburg et al. (2012): Explaining the user experience of recommender systems

* Pu, Chen and Hu (2011): A User-Centric Evaluation Framework for Recommender Systems

* Champiri et al. (2019): User Experience and Recommender Systems

10. Tourism Recommender Systems 

Recommendation for travel and tourism pose increased challenges for systems because of the complex nature of the domain. Recommender systems need to combine heterogenous data sources such as flights, hotels and attractions, take user preferences into account, but also consider context such as the best times to travel in a region. In addition, people do relatively few trips per year, so it hard to compile enough data for destination recommender systems.

  • Herzog and Wörndl (2014): A Travel Recommender System for Combining Multiple Travel Regions to a Composite Trip.
  • Ben Messaoud et al. (2017): SemCoTrip: A Variety-Seeking Model for Recommending Travel Activities in a Composite Trip 
  • Mrazovic et al. (2017): Improving Mobility in Smart Cities with Intelligent Tourist Trip Planning

11. Touristic Region Detection (Viet Duc Mai)

When recommending travel destinations from all around the globe, one needs to have a list of destinations. Often touristic regions correspond to political regions, however this not always the case. This topic is concerned how to find a touristic areas beyond political boundaries and project them on the map.

  • Dietz (2018): Data-Driven Destination Recommender Systems
  • Schlieder and Henrich (2011): Spatial grounding with vague place models
  • Adams et al. (2015): Frankenplace: Interactive Thematic Mapping for Ad Hoc Exploratory Search

12. Automatic Preference Elicitation from Social Media (Furkan Çelik)

It is said that Facebook, Google and Twitter know you better than yourself. If this was true, it would be a fruitful information source not only for targeted advertisement, but for personalized recommendations in other domains. An exemplary task would how to derive e.g., traveler types from public information of a social media profile. This could include posts, pictures and additional metadata like locations. Neidhardt et al. established the Seven Factor Model for traveler types. Is it possible to reliably derive the traveler type from a social media profile? How could this be implemented in the different social network platforms and evaluated using real persons?

  • Sertkan et al. (2018): Mapping of Tourism Destinations to Travel Behavioural Patterns
  • Neidhardt et al. (2015): A picture-based approach to recommender systems

Contact

To top