Connecting Targets to Tweets: Semantic Attention-Based Model for Target-Specific Stance Detection

Abstract

Copies of full items can be used for personal research or study, educational, or not-for profit purposes without prior permission or charge. Provided that the authors, title and full bibliographic details are credited, a hyperlink and/or URL is given for the original metadata page and the content is not changed in any way. Publisher's statement:"The final publication is available at Springer via https://doi.org/10.1007/978-3-319-68783-4_2 " A note on versions:The version presented here may differ from the published version or, version of record, if you wish to cite this item you are advised to consult the publisher's version. Please see the 'permanent WRAP url' above for details on accessing the published version and note that access may require a subscription. Abstract. Understanding what people say and really mean in tweets is still a wide open research question. In particular, understanding the stance of a tweet, which is determined not only by its content, but also by the given target, is a very recent research aim of the community. It still remains a challenge to construct a tweet's vector representation with respect to the target, especially when the target is only implicitly mentioned, or not mentioned at all in the tweet. We believe that better performance can be obtained by incorporating the information of the target into the tweet's vector representation. In this paper, we thus propose to embed a novel attention mechanism at the semantic level in the bi-directional GRU-CNN structure, which is more fine-grained than the existing token-level attention mechanism. This novel attention mechanism allows the model to automatically attend to useful semantic features of informative tokens in deciding the target-specific stance, which further results in a conditional vector representation of the tweet, with respect to the given target. We evaluate our proposed model on a recent, widely applied benchmark Stance Detection dataset from Twitter for the SemEval-2016 Task 6.A. Experimental results demonstrate that the proposed model substantially outperforms several strong baselines, which include the state-of-the-art token-level attention mechanism on bi-directional GRU outputs and the SVM classifier.

Your Stance is Exposed! Analysing Possible Factors for Stance Detection on Social Media

“…Another marginal improvement introduced by [16] by using attention based LSTM model, which achieved 68.84% F-score. Whereas in [57] the usage of bi-directional GRU-CNN yielded an F-score of 69.42%.…”

Tweet Stance Detection Using Multi-Kernel Convolution and Attentive LSTM Variants

“…Later, Du et al [19] utilized the target-augmented embeddings in an attention based neural network, whereas Zhou et al [20] proposed an attention mechanism at the semantic level in the bidirectional GRU-CNN structure to perform target-specific stance detection on tweets. More recently, Dey et al [2] proposed a two-phase LSTM based model with attention.…”

“…We compared the performance of our proposed method (MKC-LSTMVs-ATT) with the state-of-the-art stance detection methods including SemEval-2016 baselines [1], top 3 performing teams in SemEval-2016 named as MITRE [4], Pkudblab [3], and TakeLab [48], recently proposed deep learning based methods TGMN-CR [21], n-grams+embeddings [49], AS-biGRU-CNN [20], T-PAN [2], and TAN [19] as well as other related methods such as EnsembleTKPGHS [22]. The comparative results are presented in Table 4.…”

“…In the n-grams+embeddings [49] method, Mohammad et al utilized a linear SVM classifier with the word n-grams features, character n-grams features, and embedding based features drawn from additional unlabeled data. In the AS-biGRU-CNN [20] model, Zhou et al proposed a semantic level attention mechanism that attended the useful semantic features of informative tokens in a tweet. They used this attention mechanism to inject the target information into the bidirectional GRU-CNN model.…”

Untitled

“…Later, (Du et al, 2017) utilized the targetaugmented embeddings in an attention based neural network, whereas (Zhou et al, 2017) proposed an attention mechanism at the semantic level in the bidirectional GRU-CNN structure to perform target-specific stance detection on tweets. More recently, (Dey et al, 2018) proposed a two-phase LSTM based model with attention and (Wei et al, 2018b) proposed an end-to-end neural memory model via target and tweet interactions.…”