A Neural Network Approach for Truth Discovery in Social Sensing

Marshall, Jermaine; Argueta, Arturo; Wang, Dong

doi:10.1109/mass.2017.26

Cited by 13 publications

(9 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…NN [18]: This method develops a multi-layer neural network model to solve the truth discovery problem in social sensing without any assumption on the prior knowledge of the source-claim relational dependency distribution. It should be pointed that this method can only handle truth discovery problem with binary attributes.…”

Section: Experiments Protocols 1) Baseline Methodsmentioning

confidence: 99%

“…Various methods have been proposed based on the principle that the sources providing trustworthy information more often will be assigned higher reliabilities, and the information that is supported by reliable sources will be regarded as truth [19]. These truth discovery methods can be roughly divided into the following four categories: iterative methods [5]- [7], in which the truth computation step and source reliability estimation step are iteratively conducted until convergence; optimization based methods [8]- [11], in which a distance function will be defined to measure the difference between the information provided by sources and the identified truth; probabilistic graphical model based methods [12]- [14], which assumes that observations are generated based on the two parameters corresponding truth and source reliability, expectation maximization is widely used to infer the latent variables; neural network based methods [15]- [18], in which neural network is applied to accurately estimate the source-claim relational dependency function, which may be very complex.…”

Section: Related Work a Truth Discovery From Structured Datamentioning

confidence: 99%

“…Due to the characteristic of the restricted boltzmann machine, this method focus only on the data with binary attributes [21], Marshall et al used the fully connected neural network to learn the relationship between the source reliability and the claim credibility. This method also simplified the users' answers to be true or false [18]. Some researchers focused on unstructured text data, and transformed the raw text data to structured categorical data.…”

Section: B Truth Discovery From Text Datamentioning

confidence: 99%

“…A variety of methods have been proposed to estimate the source reliability and infer trustworthy information from noisy generated data [1]- [18]. Though yielding good performance, there are some drawbacks in traditional truth…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

An Unsupervised Approach of Truth Discovery From Multi-Sourced Text Data

et al. 2019

View full text Add to dashboard Cite

Truth discovery methods aim to identify which piece of information is trustworthy from multi-sourced data. Most existing truth discovery methods, however, are designed for structured data and fail to meet the strong need to extract trustworthy information from raw text data. More specifically, existing methods ignore the semantic information of text answers, i.e., answers may contain multiple factors, the word usages may be diverse, and the answers may be partially correct. In addition, ubiquitous long-tail phenomenon exists in the tasks, i.e., most users provide only a few answers and only a few users provide plenty of answers, which causes the user reliability estimation for small users to be unreasonable. To tackle these challenges, we propose a Graph Convolutional Network (GCN) based truth discovery model to automatically discover trustworthy information from text data. Firstly, Smooth Inverse Frequency (SIF) is utilized to learn real-valued vector representations for answers. Then, we construct undirected graph with these vectors to capture the structural information of answers. Finally, the GCN is utilized to store and update the reliability of these answers, and sums up all the feature vectors of all neighboring answers to improve the accuracy and efficiency of truth discovery. Different from traditional methods, we use vectors to store the reliability of answers which have higher representation capability compared with real numbers, and network is utilized to capture complex relationships among answers rather than simplified functions. The experiment results on real datasets show that though text data structures are complex, our model can still find reliable answers compared with retrieval-based and state-of-the-art truth discovery methods.INDEX TERMS Data quality, truth discovery, graph convolutional network, data mining.

show abstract

Section: Experiments Protocols 1) Baseline Methodsmentioning

confidence: 99%

Section: Related Work a Truth Discovery From Structured Datamentioning

confidence: 99%

Section: B Truth Discovery From Text Datamentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

An Unsupervised Approach of Truth Discovery From Multi-Sourced Text Data

et al. 2019

View full text Add to dashboard Cite

show abstract

“…Neural Networks. A novel neural network-based approach has been recently proposed by [14]. This method can learn complex relational dependency between source reliability and claim truthfulness.…”

Section: Novel Research Directionsmentioning

confidence: 99%