IKM at SemEval-2017 Task 8: Convolutional Neural Networks for
            stance detection and rumor verification

Chen, Yi-Chin; Kao, Hung-Yu

doi:10.18653/v1/s17-2081

Cited by 88 publications

(42 citation statements)

References 6 publications

(4 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In some cases the ensembles were hybrid, consisting both of machine learning classifiers and manually created rules, with differential weighting of classifiers for different class labels (Wang et al 2017;García Lozano et al 2017;Srivastava et al 2017). Three systems used deep learning, with Kochkina et al (2017) employing Long/Short-Term Memory Networks (LSTMs) for sequential classification; Chen et al (2017), using convolutional neural networks (CNN) for obtaining the representation of each tweet, assigned a probability for a class by a softmax classifier; and García Lozano et al (2017) using CNN as one of the classifiers in their hybrid conglomeration. The remaining two systems by Enayet and El-Beltagy (2017) and Singh et al (2017) used support vector machines with a linear and polynomical kernel, respectively.…”

Section: Approaches To Rumour Stance Classificationmentioning

confidence: 99%

“…Half of the systems invested in elaborate feature engineering, including cue words and expressions denoting belief, knowledge, doubt, and denial (Bahuleyan and Vechtomova 2017) as well as tweet domain features, including metadata about users, hashtags, and eventspecific keywords (Wang et al 2017;Bahuleyan and Vechtomova 2017;Singh et al 2017;Enayet and El-Beltagy 2017). The systems with the least-elaborate features were Chen et al (2017) and García Lozano et al (2017) for CNNs (word embeddings), Srivastava et al (2017) (sparse word vectors as input to logistic regression) and Kochkina et al (2017) (average word vectors, punctuation, similarity between word vectors in current tweet, source tweet, and previous tweet, presence of negation, picture, URL). Five out of the eight systems used pre-trained word embeddings, mostly Google News word2vec embeddings, 15 whereas Wang et al (2017) used four different types of embeddings.…”

Section: Approaches To Rumour Stance Classificationmentioning

confidence: 99%

“…Five participants submitted results to this subtask. Participants viewed the task either as a three-way (Enayet and El-Beltagy 2017;Wang et al 2017;Singh et al 2017) or two-way (Chen et al 2017;Srivastava et al 2017) single tweet classification task. The methods used mostly the same features and classifiers as used in subtask A (see Section 8), although some added features more specific to the distribution of stance labels in the tweets replying to the source tweet (for example, the best performing system in this task (Enayet and El-Beltagy 2017) considered the percentage of reply tweets classified as either support, deny, or query).…”

Section: Approaches To Rumour Veracity Classificationmentioning

confidence: 99%

See 2 more Smart Citations

Detection and Resolution of Rumours in Social Media

et al. 2018

View full text Add to dashboard Cite

Despite the increasing use of social media platforms for information and news gathering, its unmoderated nature often leads to the emergence and spread of rumours, i.e., items of information that are unverified at the time of posting. At the same time, the openness of social media platforms provides opportunities to study how users share and discuss rumours, and to explore how to automatically assess their veracity, using natural language processing and data mining techniques. In this article, we introduce and discuss two types of rumours that circulate on social media: long-standing rumours that circulate for long periods of time, and newly emerging rumours spawned during fast-paced events such as breaking news, where reports are released piecemeal and often with an unverified status in their early stages. We provide an overview of research into social media rumours with the ultimate goal of developing a rumour classification system that consists of four components: rumour detection, rumour tracking, rumour stance classification, and rumour veracity classification. We delve into the approaches presented in the scientific literature for the development of each of these four components. We summarise the efforts and achievements so far toward the development of rumour classification systems and conclude with suggestions for avenues for future research in social media mining for the detection and resolution of rumours.

show abstract

Section: Approaches To Rumour Stance Classificationmentioning

confidence: 99%

Section: Approaches To Rumour Stance Classificationmentioning

confidence: 99%

Section: Approaches To Rumour Veracity Classificationmentioning

confidence: 99%

See 1 more Smart Citation

Detection and Resolution of Rumours in Social Media

et al. 2018

View full text Add to dashboard Cite

show abstract

“…[7] used LSTM in a similar problem of early rumor detection. In an another work, [8] aimed at detecting the stance of tweets and determining the veracity of the given rumor with convolution neural networks. A submission [3] to the SemEval 2016 Twitter Stance Detection task focuses on creating a bag-of-words auto encoder, and training it over the tokenized tweets.…”

Section: Related Workmentioning

confidence: 99%

Combining Neural, Statistical and External Features for Fake News Stance Identification

Bhatt

Sharma

et al. 2018

Companion of the the Web Conference 2018 on the Web Conference 2018 - WWW '18

View full text Add to dashboard Cite

Identifying the veracity of a news article is an interesting problem while automating this process can be a challenging task. Detection of a news article as fake is still an open question as it is contingent on many factors which the current state-of-the-art models fail to incorporate. In this paper, we explore a subtask to fake news identification, and that is stance detection. Given a news article, the task is to determine the relevance of the body and its claim. We present a novel idea that combines the neural, statistical and external features to provide an efficient solution to this problem. We compute the neural embedding from the deep recurrent model, statistical features from the weighted n-gram bag-of-words model and hand crafted external features with the help of feature engineering heuristics. Finally, using deep neural layer all the features are combined, thereby classifying the headline-body news pair as agree, disagree, discuss, or unrelated. We compare our proposed technique with the current state-of-the-art models on the fake news challenge dataset. Through extensive experiments, we find that the proposed model outperforms all the state-of-the-art techniques including the submissions to the fake news challenge.

show abstract

“…Half of the systems invested in elaborate feature engineering, including cue words and expressions denoting Belief, Knowledge, Doubt and Denial [42] as well as Tweet domain features, including meta-data about users, hashtags and event specific keywords [39,42,40,45]. The systems with the least elaborate features were Chen et al [44] and García Lozano et al [41] for CNNs (word embeddings), Srivastava et al [43] (sparse word vectors as input to logistic regression) and Kochkina et al [38] (average word vectors, punctuation, similarity between word vectors in current tweet, source tweet and previous tweet, presence of negation, picture, URL). Five out of the eight systems used pre-trained word embeddings, mostly Google News word2vec embeddings 1 , whereas [39] used four different types of embeddings.…”

Section: Related Workmentioning

confidence: 99%

Discourse-aware rumour stance classification in social media using sequential classifiers

Zubiaga

Kochkina

Liakata

et al. 2018

Information Processing & Management

124

View full text Add to dashboard Cite

Rumour stance classification, defined as classifying the stance of specific social media posts into one of supporting, denying, querying or commenting on an earlier post, is becoming of increasing interest to researchers. While most previous work has focused on using individual tweets as classifier inputs, here we report on the performance of sequential classifiers that exploit the discourse features inherent in social media interactions or 'conversational threads'. Testing the effectiveness of four sequential classifiers -Hawkes Processes, Linear-Chain Conditional Random Fields (Linear CRF), Tree-Structured Conditional Random Fields (Tree CRF) and Long Short Term Memory networks (LSTM) -on eight datasets associated with breaking news stories, and looking at different types of local and contextual features, our work sheds new light on the development of accurate stance classifiers. We show that sequential classifiers that exploit the use of discourse properties in social media conversations while using only local features, outperform non-sequential classifiers. Furthermore, we show that LSTM using a reduced set of features can outperform the other sequential classifiers; this performance is consistent across datasets and across types of stances. To conclude, our work also analyses the different features under study, identifying those that best help characterise and distinguish between stances, such as supporting tweets being more likely to be accompanied by evidence than denying tweets. We also set forth a number of directions for future research.

show abstract

IKM at SemEval-2017 Task 8: Convolutional Neural Networks for stance detection and rumor verification

Cited by 88 publications

References 6 publications

Detection and Resolution of Rumours in Social Media

Detection and Resolution of Rumours in Social Media

Combining Neural, Statistical and External Features for Fake News Stance Identification

Discourse-aware rumour stance classification in social media using sequential classifiers

Contact Info

Product

Resources

About