Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

To Search or to Recommend: Predicting Open-App Motivation with Neural Hawkes Process

Zhongxiang Sun, Zihua Si, Xiao Zhang, Xiaoxue Zang, Yang Song, Hongteng Xu, Jun Xu

TL;DR

This paper tackles open-app motivation prediction in apps that combine search and recommendation by modeling temporal, multi-factor user histories. It introduces NHP-OAM, a Neural Hawkes Process-based model with a hierarchical transformer history encoder and a time-aware intensity function to predict whether the next app opening will be driven by searching or by recommendations. The authors construct the Open-App Motivation Dataset (OAMD) and extend ZhihuRec to benchmark the task, showing that NHP-OAM outperforms diverse baselines and yields meaningful improvements on downstream item recommendation and open-app ad targeting. By capturing periodicity, repeat-query behavior, and relevance features, NHP-OAM demonstrates substantial practical value for dynamically balancing S&R components and enhancing user engagement. The work provides open datasets, a rigorous modeling framework, and evidence of real-world impact in S&R-enabled applications.

Abstract

Incorporating Search and Recommendation (S&R) services within a singular application is prevalent in online platforms, leading to a new task termed open-app motivation prediction, which aims to predict whether users initiate the application with the specific intent of information searching, or to explore recommended content for entertainment. Studies have shown that predicting users' motivation to open an app can help to improve user engagement and enhance performance in various downstream tasks. However, accurately predicting open-app motivation is not trivial, as it is influenced by user-specific factors, search queries, clicked items, as well as their temporal occurrences. Furthermore, these activities occur sequentially and exhibit intricate temporal dependencies. Inspired by the success of the Neural Hawkes Process (NHP) in modeling temporal dependencies in sequences, this paper proposes a novel neural Hawkes process model to capture the temporal dependencies between historical user browsing and querying actions. The model, referred to as Neural Hawkes Process-based Open-App Motivation prediction model (NHP-OAM), employs a hierarchical transformer and a novel intensity function to encode multiple factors, and open-app motivation prediction layer to integrate time and user-specific information for predicting users' open-app motivations. To demonstrate the superiority of our NHP-OAM model and construct a benchmark for the Open-App Motivation Prediction task, we not only extend the public S&R dataset ZhihuRec but also construct a new real-world Open-App Motivation Dataset (OAMD). Experiments on these two datasets validate NHP-OAM's superiority over baseline models. Further downstream application experiments demonstrate NHP-OAM's effectiveness in predicting users' Open-App Motivation, highlighting the immense application value of NHP-OAM.

To Search or to Recommend: Predicting Open-App Motivation with Neural Hawkes Process

TL;DR

This paper tackles open-app motivation prediction in apps that combine search and recommendation by modeling temporal, multi-factor user histories. It introduces NHP-OAM, a Neural Hawkes Process-based model with a hierarchical transformer history encoder and a time-aware intensity function to predict whether the next app opening will be driven by searching or by recommendations. The authors construct the Open-App Motivation Dataset (OAMD) and extend ZhihuRec to benchmark the task, showing that NHP-OAM outperforms diverse baselines and yields meaningful improvements on downstream item recommendation and open-app ad targeting. By capturing periodicity, repeat-query behavior, and relevance features, NHP-OAM demonstrates substantial practical value for dynamically balancing S&R components and enhancing user engagement. The work provides open datasets, a rigorous modeling framework, and evidence of real-world impact in S&R-enabled applications.

Abstract

Incorporating Search and Recommendation (S&R) services within a singular application is prevalent in online platforms, leading to a new task termed open-app motivation prediction, which aims to predict whether users initiate the application with the specific intent of information searching, or to explore recommended content for entertainment. Studies have shown that predicting users' motivation to open an app can help to improve user engagement and enhance performance in various downstream tasks. However, accurately predicting open-app motivation is not trivial, as it is influenced by user-specific factors, search queries, clicked items, as well as their temporal occurrences. Furthermore, these activities occur sequentially and exhibit intricate temporal dependencies. Inspired by the success of the Neural Hawkes Process (NHP) in modeling temporal dependencies in sequences, this paper proposes a novel neural Hawkes process model to capture the temporal dependencies between historical user browsing and querying actions. The model, referred to as Neural Hawkes Process-based Open-App Motivation prediction model (NHP-OAM), employs a hierarchical transformer and a novel intensity function to encode multiple factors, and open-app motivation prediction layer to integrate time and user-specific information for predicting users' open-app motivations. To demonstrate the superiority of our NHP-OAM model and construct a benchmark for the Open-App Motivation Prediction task, we not only extend the public S&R dataset ZhihuRec but also construct a new real-world Open-App Motivation Dataset (OAMD). Experiments on these two datasets validate NHP-OAM's superiority over baseline models. Further downstream application experiments demonstrate NHP-OAM's effectiveness in predicting users' Open-App Motivation, highlighting the immense application value of NHP-OAM.
Paper Structure (31 sections, 20 equations, 7 figures, 3 tables)

This paper contains 31 sections, 20 equations, 7 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Empirical study
  4. Task Definition
  5. Dataset Description
  6. Behavioral Patterns Behind Users' Open-App Motivation
  7. Periodicity in Open-App Motivation
  8. Repeat-Query in Open-App Motivation
  9. Relevance in Open-App Motivation
  10. Our Approach: NHP-OAM
  11. Preliminaries about Neural Hawkes Process
  12. NHP-OAM: History Encoder
  13. Embedding Layer.
  14. Session-level Encoder.
  15. History-level Encoder.
  16. ...and 16 more sections

Figures (7)

  • Figure 1: Examples of Open-App Motivation. Obtained through a structured questionnaire-based interview process with users.
  • Figure 2: Periodicity statistics in Open-App Motivation. OAM: S denotes the Open-App Motivation: Search. OAM: R denotes the Open-App Motivation: Recommendation.
  • Figure 3: Open-App Motivation Features. (a) Repeat-query statistics in open-app motivation to search. Users are divided into four activity groups by Open-App Motivate: Search (OAM: S) activity levels (high values refer to high activity levels). The overall average denotes the repeat-query ratio in all the users. (b) Relevance statistics between the ratio of clicked search list to clicked recommended list in users' past sessions and their motivation to search when they open the APP next time.
  • Figure 4: The overall architecture of the proposed model NHP-OAM, featuring Session-level Encoder and History-level Encoder in § \ref{['sec:History_Encoder']}, Type-specific Intensity Function in § \ref{['sec:Intensity_Function']} and Open-App Motivation Prediction Layer in § \ref{['sec:Open-App Motivation Prediction Layer']}.
  • Figure 5: Application on Item Recommendation.
  • ...and 2 more figures