Taxonomy Completion via Triplet Matching Network

Zhang, Jieyu; Song, Xiangchen; Zeng, Ying; Chen, Jiaze; Shen, Jiaming; Mao, Yuning; Li, Lei

doi:10.48550/arxiv.2101.01896

Cited by 3 publications

(8 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, for every node in the existing taxonomy, all concepts in C can be taken as candidates of hyponym, thus the potential search space is incredibly enormous. Previous studies [16]- [19], [23] address the above problem via an assumption that the input set of new concepts contains only one element (i.e., |C| = 1), and aims to find one single parent node in the existing taxonomy T 0 of this new concept (i.e., |R| = 1). This simplification not only limits expansion in the depth of taxonomies, but also neglects the truth that a new concept has multiple parent nodes.…”

Section: B Problem Definitionmentioning

confidence: 99%

“…Prior studies like [58], [59] require manual training data to learn features. Self-supervised constructing training set from existing taxonomy is a common strategy considered to address this problem in recent works [16], [18], [19], [23]. In our work, we further propose an adaptively self-supervised generation strategy to avoid inheriting the adverse problems in the existing taxonomy.…”

Section: B Taxonomy Expansionmentioning

confidence: 99%

See 1 more Smart Citation

Learning What You Need from What You Did: Product Taxonomy Expansion with User Behaviors Supervision

Cheng¹,

Gu²,

Liu³

et al. 2022

Preprint

View full text Add to dashboard Cite

Taxonomies have been widely used in various domains to underpin numerous applications. Specially, product taxonomies serve an essential role in the e-commerce domain for the recommendation, browsing, and query understanding. However, taxonomies need to constantly capture the newly emerged terms or concepts in e-commerce platforms to keep up-to-date, which is expensive and labor-intensive if it relies on manual maintenance and updates. Therefore, we target the taxonomy expansion task to attach new concepts to existing taxonomies automatically. In this paper, we present a self-supervised and user behavior-oriented product taxonomy expansion framework to append new concepts into existing taxonomies. Our framework extracts hyponymy relations that conform to users' intentions and cognition. Specifically, i) to fully exploit user behavioral information, we extract candidate hyponymy relations that match user interests from query-click concepts; ii) to enhance the semantic information of new concepts and better detect hyponymy relations, we model concepts and relations through both usergenerated content and structural information in existing taxonomies and user click logs, by leveraging Pre-trained Language Models and Graph Neural Network combined with Contrastive Learning; iii) to reduce the cost of dataset construction and overcome data skews, we construct a high-quality and balanced training dataset from existing taxonomy with no supervision. Extensive experiments on real-world product taxonomies in Meituan Platform, a leading Chinese vertical e-commerce platform to order take-out with more than 70 million daily active users, demonstrate the superiority of our proposed framework over state-of-the-art methods. Notably, our method enlarges the size of real-world product taxonomies from 39,263 to 94,698 relations with 88% precision. Our implementation is available: https://github.com/AdaCheng/Product Taxonomy Expansion.

show abstract

Section: B Problem Definitionmentioning

confidence: 99%

Section: B Taxonomy Expansionmentioning

confidence: 99%

Learning What You Need from What You Did: Product Taxonomy Expansion with User Behaviors Supervision

Cheng¹,

Gu²,

Liu³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Figure 3 illustrates an example of an ego tree. A similar approach has also been used in (Wang et al, 2021;Yu et al, 2020b;Zhang et al, 2021;Shen et al, 2020) studies. To obtain the embedding of the anchor concept, we further apply graph embedding as described below.…”

Section: Team-regression (Team-rg)mentioning

confidence: 99%

“…Though both of these operations are integral parts of a WordNet taxonomy expansion, all of the existing studies on taxonomy expansion have considered expansion with either attach operation (Schlichtkrull and Alonso, 2016;Vedula et al, 2018;Shen et al, 2020;Yu et al, 2020b;Zhang et al, 2021;Takeoka et al, 2021; or merge operation (Nakashole et al, 2012;Nguyen et al, 2017;Nakashole et al, 2012;Qu et al, 2017;Boteanu et al, 2018;Wang et al, 2019;Fei et al, 2019), but not together. Realizing the need to apply both the operation, SemEval-2016:task 14 (Semantic taxonomy enrichment) Jurgens and Pilehvar (2016) includes a call for expansion with both attach and merge operations.…”

Section: Introductionmentioning

confidence: 99%

“…Motivated by the above observations, in this study, we propose an integrated deep learningbased method, namely, Taxonomy Expansion with Attach and Merge (TEAM), which performs both the attach and merge operations in a multitasklearning framework. Though most of the existing studies consider the expansion a regression problem (Shen et al, 2020;Yu et al, 2020b;Zhang et al, 2021), considering that our method performs both the attach and merge operation in a single model, it can also be considered a classification task. As a result, we propose two versions of TEAM, namely, TEAM-RG: Regression, and TEAM-CL: Classification to perform with explicit and implicit rankings.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

TEAM: A multitask learning based Taxonomy Expansion approach for Attach and Merge

Phukon¹,

Mitra²,

Sanasam³

et al. 2022

Findings of the Association for Computational Linguistics: NAACL 2022

View full text Add to dashboard Cite

Taxonomy expansion is a crucial task. Most of the taxonomy expansion approaches are of two types, attach and merge. In a taxonomy like WordNet, both merge and attach are integral parts of the expansion operations, but the majority of studies consider them separately. This paper proposes a novel multi-task learning-based deep learning method known as Taxonomy Expansion with Attach and Merge (TEAM) that performs both the merge and attach operations. This is the first study that integrates both the merge and attach operations in a single model to the best of our knowledge. The proposed models have been evaluated on three separate WordNet taxonomies, viz., Assamese, Bangla, and Hindi. From the various experimental setups, it is shown that TEAM outperforms its state-of-the-art counterparts for attach operation and also provides highly encouraging performance for the merge operation.

show abstract

GANTEE: Generative Adversarial Network for Taxonomy Enterance Evaluation

Jiang

Liu

et al. 2023

AAAI

View full text Add to dashboard Cite

Taxonomy is formulated as directed acyclic graphs or trees of concepts that support many downstream tasks. Many new coming concepts need to be added to an existing taxonomy. The traditional taxonomy expansion task aims only at finding the best position for new coming concepts in the existing taxonomy. However, they have two drawbacks when being applied to the real-scenarios. The previous methods suffer from low-efficiency since they waste much time when most of the new coming concepts are indeed noisy concepts. They also suffer from low-effectiveness since they collect training samples only from the existing taxonomy, which limits the ability of the model to mine more hypernym-hyponym relationships among real concepts. This paper proposes a pluggable framework called Generative Adversarial Network for Taxonomy Entering Evaluation (GANTEE) to alleviate these drawbacks. A generative adversarial network is designed in this framework by discriminative models to alleviate the first drawback and the generative model to alleviate the second drawback. Two discriminators are used in GANTEE to provide long-term and short-term rewards, respectively. Moreover, to further improve the efficiency, pre-trained language models are used to retrieve the representation of the concepts quickly. The experiments on three real-world large-scale datasets with two different languages show that GANTEE improves the performance of the existing taxonomy expansion methods in both effectiveness and efficiency.

show abstract

Taxonomy Completion via Triplet Matching Network

Cited by 3 publications

References 30 publications

Learning What You Need from What You Did: Product Taxonomy Expansion with User Behaviors Supervision

Learning What You Need from What You Did: Product Taxonomy Expansion with User Behaviors Supervision

TEAM: A multitask learning based Taxonomy Expansion approach for Attach and Merge

GANTEE: Generative Adversarial Network for Taxonomy Enterance Evaluation

Contact Info

Product

Resources

About