icon

2019-07-13

User Modeling, Adaptation, Personalization for Marketing #UMAP2019

It has been a month since I presented at The 27th ACM Conference On User Modeling, Adaptation And Personalization (UMAP2019) held in Larnaca, Cyprus:

* You can download the PDF file from HERE

You may also be interested in a following video I created during the submission process of the demo paper:

UMAP in Customer Data Platform

The accepted paper has demonstrated UMAP capabilities implemented in our Arm Treasure Data Customer Data Platform with a poster:

* PDF version of the poster is available at HERE

Customer Data Platform (CDP) is a new type of marketer-operated applications that allows the marketing team to obtain a deeper understanding of individual customers.

Unlike traditional Customer Relationship Management (CRM) systems, CDP application gives a more detailed view of the customers by unifying not only their static attributes (e.g., name, email address) but dynamic behaviors coming from a wide variety of online and offline data sources such as web access logs, third-party sales/marketing tools, on-premise DBs, in-store physical sensors.

In the demo session, one of the most frequently asked questions was: What is a relationship between UMAP and CDP?

Well, first of all, CDP itself is a tool for Adaptation and Personalization. Importantly, the reason why marketers try to understand more about you (as a customer) is to provide better personalized experiences, and CDP makes the efforts easy and simple. As a consequence of personalized digital and physical customer communication, marketers can effectively and efficiently use their financial, human, and physical resources.

Meanwhile, since real-world customer data is exceptionally complicated, User Modeling techniques play a really important role in the practical CDP implementation. For instance, if marketers can accurately predict customer's future conversion, they can easily optimize (and even automate) their day-to-day marketing activities to quickly achieve Adaptation and Personalization. Even if the technique is not a state-of-the-art one, it still has a huge business impact compared to the traditional "manual" way of marketing.

For the reason that I mentioned above, I argued the strong relationship between UMAP and CDP in the paper, by using an example of our commercially-available CDP implementation.

Trend: Transparency and Explainability

We can confidently say that transparency and explainability of user modeling are currently one of the most important topics in this area. Many talks I attended similarly mentioned the importance, and the theme of this year's UMAP conference "Making Personalization Transparent: Giving Control Back To The User" also indicates the fact. The trend does make sense for the following reasons.

First, as a developer of modern UMAP applications, we highly rely on machine learning techniques to capture the characteristics of data. However, at the same time, the way of user modeling is getting more and more complicated as the AI research makes progress. Hence, explainability helps us a lot to understand why algorithms made this prediction, and eventually we may be able to get new insights from the explanations.

More importantly, since end users of UMAP applications are usually unfamiliar with underlying technical details, the developers are required to provide a certain degree of explanation and make the things transparent to ensure better user experiences. Otherwise, users can be surprised when they receive "unexpected" (e.g., inaccurate) personalization for some reasons, and it naturally makes people uncomfortable. In order to minimize the risk, our enterprise CDP internally uses very simple techniques such as logistic regression and TF-IDF weighting to keep the balance between simplicity, accuracy, and explainability, for example.

Therefore, transparency and explainability are really important in the UMAP context. Note that I saw a similar trend in the field of recommender systems, which is actually a subset of UMAP studies, when I attended last year's RecSys conference.

Days in Cyprus

In short, Cyprus is a beautiful island where makes you relax. While I attended the conference alone, I really enjoyed the weather, drinks, foods, people, and the beach during my stay.

Unfortunately, I needed to skip a majority of the conference program for work, but I can still say UMAP2019 was worthy to attend. I learned a lot from the limited, but informative talks I heard, and I met nice people coming from academia and industry across many different countries.

I would like to thank the organizers and attendees for making the conference great.

search See also

» more
user Takuya Kitazawa (a.k.a. takuti) is an engineer working on machine learning, data science, and product development at Arm Treasure Data. Opinions are my own.