A Three-Step Method for Audience Extension in Internet Advertising Using an Industrial Taxonomy

Dmitry Frolov and Zina Taran
Additional contact information
Dmitry Frolov: HSE University
Zina Taran: Delta State University

A chapter in Data Analysis and Optimization, 2023, pp 135-146 from Springer

Abstract: Abstract The paper addresses a very common problem in targeted digital advertising, insufficient audience size. Many approaches to audience extension frequently lead to much diminishing quality metrics, such as audience quality or conversion rates. This is the case, for example, for so-called look-alike techniques. We present a novel method for the efficient extension of target audiences. Our base is a popular taxonomy of user interests, the IAB Contents taxonomy, combined with the representation of browsing behavior of millions of users by fuzzy sets of visited IAB taxonomy segments, that are leaves of the taxonomy tree. We use this idea in our method. The method consists of three steps: (1) computing membership values for the interest segments for a user by a classifier; (2) performing generalization of those sets and obtaining high-ranked segments, which is a core part of the method; (3) obtaining a set of advertising campaigns for a user. Our method involves an algorithm for optimally lifting individual fuzzy leaf sets into a higher rank taxonomy node, a so-called “head subject”. The head subject must cover the input fuzzy leaf set in such a way that the number of errors is minimized. This algorithm was proposed as an intelligent information retrieval tool. It can be applied, however, to a very different task of targeted advertisement. To extend the audiences of a targeted advertisement, we find their head subjects off-line. Given a set of taxonomy segments corresponding to targeted audiences, we include a user as a target if her head subject covers any of those segments. This lifting-based step does increase the number of successful matches between user segments and campaign segments two- or three-fold without losing in the targeting quality, the click-through rate, CTR. This is in stark contrast to the conventional look-alike methods for increasing the audience numbers by reducing the admissibility thresholds, which leads to a large decrease in CTR, which was experimentally proven.

Keywords: Targeted advertising; Audience extension; Taxonomy; Fuzzy leaf set; Optimal lifting (search for similar items in EconPapers)
Date: 2023
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:spochp:978-3-031-31654-8_8

Ordering information: This item can be ordered from
http://www.springer.com/9783031316548

DOI: 10.1007/978-3-031-31654-8_8

Access Statistics for this chapter

More chapters in Springer Optimization and Its Applications from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().