Aspect terms extraction (ATE) and aspect sentiment classification (ASC) are two fundamental and fine-grained sub-tasks in aspect-level sentiment analysis (ALSA). In the textual analysis, joint extracting both aspect terms and sentiment polarities has
been drawn much attention due to the better applications than individual sub-task. However, in the multi-modal scenario, the existing studies are limited to handle each sub-task independently, which fails to model the innate connection between the above two objectives and ignores the better applications. Therefore, in this paper, we are the first to jointly perform multi-modal ATE (MATE) and multi-modal ASC (MASC), and we propose a multi-modal joint learning approach with auxiliary cross-modal relation detection for multi-modal aspect-level sentiment analysis (MALSA). Specifically, we first build an auxiliary text-image relation detection module to control the proper exploitation of visual information. Second, we adopt the hierarchical framework to bridge the multi-modal connection between MATE and MASC, as well as separately visual guiding for each sub module. Finally, we can obtain all aspect-level sentiment polarities dependent on the jointly extracted specific aspects. Extensive experiments show the effectiveness of our approach against the joint textual approaches, pipeline and collapsed multi-modal approaches.
Despite their success, modern language models are fragile. Even small changes in their training pipeline can lead to unexpected results. We study this phenomenon by examining the robustness of ALBERT (Lan et al., 2020) in combination with Stochastic
Weight Averaging (SWA)---a cheap way of ensembling---on a sentiment analysis task (SST-2). In particular, we analyze SWA's stability via CheckList criteria (Ribeiro et al., 2020), examining the agreement on errors made by models differing only in their random seed. We hypothesize that SWA is more stable because it ensembles model snapshots taken along the gradient descent trajectory. We quantify stability by comparing the models' mistakes with Fleiss' Kappa (Fleiss, 1971) and overlap ratio scores. We find that SWA reduces error rates in general; yet the models still suffer from their own distinct biases (according to CheckList).
Multimodal sentiment analysis (MSA) draws increasing attention with the availability of multimodal data. The boost in performance of MSA models is mainly hindered by two problems. On the one hand, recent MSA works mostly focus on learning cross-modal
dynamics, but neglect to explore an optimal solution for unimodal networks, which determines the lower limit of MSA models. On the other hand, noisy information hidden in each modality interferes the learning of correct cross-modal dynamics. To address the above-mentioned problems, we propose a novel MSA framework Modulation Model for Multimodal Sentiment Analysis (M3SA) to identify the contribution of modalities and reduce the impact of noisy information, so as to better learn unimodal and cross-modal dynamics. Specifically, modulation loss is designed to modulate the loss contribution based on the confidence of individual modalities in each utterance, so as to explore an optimal update solution for each unimodal network. Besides, contrary to most existing works which fail to explicitly filter out noisy information, we devise a modality filter module to identify and filter out modality noise for the learning of correct cross-modal embedding. Extensive experiments on publicly datasets demonstrate that our approach achieves state-of-the-art performance.
This technical report aims at the ROCLING 2021 Shared Task: Dimensional Sentiment Analysis for Educational Texts. In order to predict the affective states of Chinese educational texts, we present a practical framework by employing pre-trained languag
e models, such as BERT and MacBERT. Several valuable observations and analyses can be drawn from a series of experiments. From the results, we find that MacBERT-based methods can deliver better results than BERT-based methods on the verification set. Therefore, we average the prediction results of several models obtained using different settings as the final output.
In this shared task, this paper proposes a method to combine the BERT-based word vector model and the LSTM prediction model to predict the Valence and Arousal values in the text. Among them, the BERT-based word vector is 768-dimensional, and each wor
d vector in the sentence is sequentially fed to the LSTM model for prediction. The experimental results show that the performance of our proposed method is better than the results of the Lasso Regression model.
Aspect Category Sentiment Analysis (ACSA), which aims to identify fine-grained sentiment polarities of the aspect categories discussed in user reviews. ACSA is challenging and costly when conducting it into real-world applications, that mainly due to
the following reasons: 1.) Labeling the fine-grained ACSA data is often labor-intensive. 2.) The aspect categories will be dynamically updated and adjusted with the development of application scenarios, which means that the data must be relabeled frequently. 3.) Due to the increase of aspect categories, the model must be retrained frequently to fast adapt to the newly added aspect category data. To overcome the above-mentioned problems, we introduce a novel Meta Multi-Task Learning (MMTL) approach, that frame ACSA tasks as a meta-learning problem (i.e., regarding aspect-category sentiment polarity classification problems as the different training tasks for meta-learning) to learn an ideal and shareable initialization for the multi-task learning model that can be adapted to new ACSA tasks efficiently and effectively. Experiment results show that the proposed approach significantly outperforms the strong pre-trained transformer-based baseline model, especially, in the case of less labeled fine-grained training data.
Sentiment analysis aims to detect the overall sentiment, i.e., the polarity of a sentence, paragraph, or text span, without considering the entities mentioned and their aspects. Aspect-based sentiment analysis aims to extract the aspects of the given
target entities and their respective sentiments. Prior works formulate this as a sequence tagging problem or solve this task using a span-based extract-then-classify framework where first all the opinion targets are extracted from the sentence, and then with the help of span representations, the targets are classified as positive, negative, or neutral. The sequence tagging problem suffers from issues like sentiment inconsistency and colossal search space. Whereas, Span-based extract-then-classify framework suffers from issues such as half-word coverage and overlapping spans. To overcome this, we propose a similar span-based extract-then-classify framework with a novel and improved heuristic. Experiments on the three benchmark datasets (Restaurant14, Laptop14, Restaurant15) show our model consistently outperforms the current state-of-the-art. Moreover, we also present a novel supervised movie reviews dataset (Movie20) and a pseudo-labeled movie reviews dataset (moviesLarge) made explicitly for this task and report the results on the novel Movie20 dataset as well.
Recently, the majority of sentiment analysis researchers focus on target-based sentiment analysis because it delivers in-depth analysis with more accurate results as compared to traditional sentiment analysis. In this paper, we propose an interactive
learning approach to tackle a target-based sentiment analysis task for the Arabic language. The proposed IA-LSTM model uses an interactive attention-based mechanism to force the model to focus on different parts (targets) of a sentence. We investigate the ability to use targets, right, and left context, and model them separately to learn their own representations via interactive modeling. We evaluated our model on two different datasets: Arabic hotel review and Arabic book review datasets. The results demonstrate the effectiveness of using this interactive modeling technique for the Arabic target-based task. The model obtained accuracy values of 83.10 compared to SOTA models such as AB-LSTM-PC which obtained 82.60 for the same dataset.
Opinion target extraction and opinion term extraction are two fundamental tasks in Aspect Based Sentiment Analysis (ABSA). Many recent works on ABSA focus on Target-oriented Opinion Words (or Terms) Extraction (TOWE), which aims at extracting the cor
responding opinion words for a given opinion target. TOWE can be further applied to Aspect-Opinion Pair Extraction (AOPE) which aims at extracting aspects (i.e., opinion targets) and opinion terms in pairs. In this paper, we propose Target-Specified sequence labeling with Multi-head Self-Attention (TSMSA) for TOWE, in which any pre-trained language model with multi-head self-attention can be integrated conveniently. As a case study, we also develop a Multi-Task structure named MT-TSMSA for AOPE by combining our TSMSA with an aspect and opinion term extraction module. Experimental results indicate that TSMSA outperforms the benchmark methods on TOWE significantly; meanwhile, the performance of MT-TSMSA is similar or even better than state-of-the-art AOPE baseline models.
The prominence of figurative language devices, such as sarcasm and irony, poses serious challenges for Arabic Sentiment Analysis (SA). While previous research works tackle SA and sarcasm detection separately, this paper introduces an end-to-end deep
Multi-Task Learning (MTL) model, allowing knowledge interaction between the two tasks. Our MTL model's architecture consists of a Bidirectional Encoder Representation from Transformers (BERT) model, a multi-task attention interaction module, and two task classifiers. The overall obtained results show that our proposed model outperforms its single-task and MTL counterparts on both sarcasm and sentiment detection subtasks.
