Zero-Shot Heterogeneous Transfer Learning from Recommender Systems to Cold-Start Search Retrieval

Abstract

Many recent advances in neural information retrieval models, which predict top-items given a query, learn directly from a large training set of (query, item) pairs. However, they are often insufficient when there are many previously unseen (query, item) combinations, often referred to as the cold start problem. Furthermore, the search system can be biased towards items that are frequently shown to a query previously, also known as the "rich get richer" (a.k.a. feedback loop) problem. In light of these problems, we observed that most online content platforms have both a search and a recommender system that, while having heterogeneous input spaces, can be connected through their common output item space and a shared semantic representation. In this paper, we propose a new Zero-Shot Heterogeneous Transfer Learning framework that transfers learned knowledge from the recommender system component to improve the search component of a content platform. First, it learns representations of items and their natural-language features by predicting (item, item) correlation graphs derived from the recommender system as an auxiliary task. Then, the learned representations are transferred to solve the target search retrieval task, performing query-to-item prediction without having seen any (query, item) pairs in training. We conduct online and offline experiments on one of the world's largest search and recommender systems from Google, and present the results and lessons learned. We demonstrate that the proposed approach can achieve high performance on offline search retrieval tasks, and more importantly, achieved significant improvements on relevance and user interactions over the highly-optimized production system in online experiments.

Semantic-aware Retrieval and Recommendation based on the Dirichlet Compound Language Model

“…age group, language and genre for movies). There has been much interest in solving cold-start in recent papers, for example [55] proposed Zero-Shot Heterogeneous Transfer Learning in which the cold-start problem (no user-item data) is answered by predicting (item, item) correlation graphs. Also, [56] presented a content-aware neural hashing-based collaborative filtering approach (NeuHash-CF), which generates binary hash codes for users and items.…”

A Model-Agnostic Causal Learning Framework for Recommendation using Search Data

“…Besides the joint modeling, methods also have been developed to make use of search or recommendation as the external information to improve the performances of recommendation and search [31,34,39]. Wu et al [34] propose a Zero-Shot Heterogeneous Transfer Learning framework that transfers the learned knowledge from the recommendation component to the search component, addressed the cold-start problem in the search system. Wu et al [31], Yao et al [39] use the search history log to enhance the recommendation task as external information.…”

CausPref: Causal Preference Learning for Out-of-Distribution Recommendation

“…However, they lack the general formulation for distribution shift, hence can only address the bias of specific form in transductive setting. Moreover, the requirement on prior knowledge of test environment limits the transfer learning-based [41] and RL/bandit-based [42] methods to deal with the agnostic distribution shift. To pursuit the stable performance for agnostic distribution, in this paper we propose to capture the essential invariant mechanism from user behavior data to solve the common distribution shift problems, and design a model based on available features (referring to the literature of cold start [14]) because the OOD recommendation usually accompanies the emergency of new users/items.…”