Weakly-supervised Domain Adaption for Aspect Extraction via Multi-level Interaction Transfer

TL;DR

This work conducts a pioneer study on leveraging sentence-level aspect category labels that can be usually available in commercial services like review sites to promote token-level transfer for the extraction purpose and proposes a novel multi-level reconstruction mechanism that aligns both the fine- grained and coarse-grained information in multiple levels of abstractions.

Abstract

Fine-grained aspect extraction is an essential sub-task in aspect based opinion analysis. It aims to identify the aspect terms (a.k.a. opinion targets) of a product or service in each sentence. However, expensive annotation process is usually involved to acquire sufficient token-level labels for each domain. To address this limitation, some previous works propose domain adaptation strategies to transfer knowledge from a sufficiently labeled source domain to unlabeled target domains. But due to both the difficulty of fine-grained prediction problems and the large domain gap between domains, the performance remains unsatisfactory. This work conducts a pioneer study on leveraging sentence-level aspect category labels that can be usually available in commercial services like review sites to promote token-level transfer for the extraction purpose. Specifically, the aspect category information is used to construct pivot knowledge for transfer with assumption that the interactions between sentence-level aspect category and token-level aspect terms are invariant across domains. To this end, we propose a novel multi-level reconstruction mechanism that aligns both the fine-grained and coarse-grained information in multiple levels of abstractions. Comprehensive experiments demonstrate that our approach can fully utilize sentence-level aspect category labels to improve cross-domain aspect extraction with a large performance gain.

Figures (4)

• Figure 1: The overall architecture of our model. The aspect extraction module produces the token-level aspect term labels. The sentence categorization module produces the sentence-level aspect category labels. The interaction transfer module contains multiple reconstruction units to associate the token-level and sentence-level information. The sentence categorization module and the aspect extraction module share a common multi-layer Bi-LSTM for transforming input word vectors $\mathbf{w}$ to high-level context-sensitive features $\mathbf{h}^T$ and a multi-head attention layer for encoding complex word dependencies to produce hidden vector $\mathbf{h}$.
• Figure 2: Visualization of attention weights $\alpha_{i}^{g}$ for different tokens within a sequence. The true aspect tokens are displayed in bold text.
• Figure 3: Sensitivity analysis for trade-off parameters. The sensitivity analysis is conducted through changing the value of $\lambda$ ($\beta$), while fixing $\beta$ ($\lambda$) to the value used in the experiments.
• Figure 4: F1 score vs proportion of target training data. The horizontal axis indicates the proportion of unlabeled target data.