News is a medium that notifies people about the events that had happened worldwide. The menace of fake news on online platforms is on the rise which may lead to unwanted events. The majority of fake news is spread through social media platforms, since these platforms have a great reach. To identify the credibility of the news, various spam detection methods are generally used. In this work, a new stance detection method has been proposed for identifying the stance of fake news. The proposed stance detection method is based on the capabilities of an improved whale optimization algorithm and a multilayer perceptron. In the proposed model, weights and biases of the multilayer perceptron are updated using an improved whale optimization algorithm. The efficacy of the proposed optimized neural network has been tested on five benchmark stance detection datasets. The proposed model shows better results over all the considered datasets. The proposed approach has theoretical implications for further studies to examine the textual data. Besides, the proposed method also has practical implications for developing systems that can result conclusive reviews on any social problems.
Hinweise
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Introduction
News is a medium that keeps everyone updated about the events that have taken place. It has two parts, one is the headline of the news and the other is the content body. News has the highest reach among all other forms of media [88]. In this era of digitization, news also spread through various social media platforms such as Facebook, Twitter, Whatsapp, etc. Nowadays, doubt regarding the credibility of news has become widespread, since some miscreates are spreading fake news purposefully through different social media platforms. Fake news can have a severe threat to the community due to its extensive reach [86]. In this digital era, everyday terabytes of fake news are created and shared on various online platforms. However, such a high-dimensional data cannot be managed by human beings in real time and human fact-checkers cannot handle such tremendous information in real time. Thus, to automate the process of fake news detection, several artificial intelligence (AI)-based techniques have been introduced in the literature. One of the popular approach to discover fake news is by identifying the stance of the news [50]. Stance detection involves estimating the relative views (or stance) of two chunks of text relative to a claim, issue, or topic. In general, a stance is defined as a relationship between two textual bodies. In stance detection, a headline and a body text are given and the objective is to classify the headline–body pair into one of the categories, namely, agree, disagree, discuss, and unrelated as discussed below.
1.
Agree stance A stance is agree stance when the claim of one body is validated by the other body. For example: Body 1-Mango is known as king of fruits.Body 2-Mango has a great taste, that is why it is known as king of fruits.
2.
Disagree stance A stance is disagree stance when the claim of one body is denied by the other body. The same is elaborated in following example: Body 1-Mango is known as king of fruits.Body 2-Mango has a great taste but we cannot say it as king of fruits.
3.
Discuss stance A stance is discuss stance when the claim of one body is neither denied nor validated by the other body rather the other body discuss the claim made by the other body. For example: Body 1-Mango is known as king of fruits.Body 2-Mango has a great taste and has great texture which makes him stand apart from the other fruits.
4.
Unrelated stance A stance is an unrelated stance when the claim of one body is neither denied nor validated by the other body rather the other body discuss anything apart form the claim made by the other body. For example: Body 1-Mango is known as king of fruits.Body 2-Arvind Kejriwal is the chief minister of the Delhi.
To discover whether a piece of particular news is fake or not, its stance is identified [32]. If the stance of news belongs to an “agreed” category, then it confirms its genuinity, while, if the stance is “‘disagreed” or “unrelated”, it implies that the news is fake. It is very difficult to discover whether the news is fake or not if the stance belongs to the discussed category. News belonging to the “discussed” category have to be physically analyzed. To discover the stance of fake news, many models such as traditional stance, machine learning, deep learning, and natural language processing (NLP)-based models have been proposed in the literature. The traditional stance-based models compare the body and headlines to check the credibility of news [27]. Shu et al. [71] presented an NLP-based approach for identifying the stance of fake news. The NLP-based approaches investigate fake news from textual and network perspective. The NLP-based approaches cannot capture semantics from textual data; hence, these approaches sometimes do not perform well. Therefore, various word embeddings such as word2vec, Glove, etc., along with machine learning are used for identifying the stance [76]. Ghanem et al. [28] introduced a hybrid model based on the strength of n-grams, lexical representation of indicative words, and word embeddings for stance detection of fake news. In the literature, it has been stated that the performance of stance classification can be improved if appropriate representative features are used. However, the existing approaches use some predefined lexicons and word embeddings to extract features from textual data. Thus, it is possible that the extracted features may be irrelevant. To extract relevant features, deep learning models are generally used.
The deep neural network-based models extract useful features automatically from datasets using backpropagation [63]. However, the performance of backpropagation falls rapidly if the complexity of the problem increases [58]. Moreover, the efficiency of neural network models also depends upon hyper-parameters such as filter window size, learning rate, word embedding techniques, and batch size. Therefore, to improve the performance and optimize the values of hyper-parameters, metaheuristic methods are generally used. The whale optimization algorithm (WOA) is successfully used in literature for optimizing FFNNs [5]. Moreover, WOA is also used to solve text classification and sentiment analysis problems [49, 78]. In this paper, an improved whale optimization algorithm (IWOA) has been proposed for optimizing the hyper-parameters of neural networks. The proposed model uses the word embedding technique for normalizing the textual data followed by an optimized neural network to get the stance. The proposed IWOA optimized FFNN can be used for solving single-objective minimization and maximization optimization problems. Furthermore, the proposed algorithm can also be extended to solve multi-objective optimization problems such as allocating parking lots in a distribution network, optimization of RO desalination plants, etc. The main contributions of this article can be outlined as follows.
A new variant of whale optimization has been proposed and validated on standard benchmark functions.
The proposed improved whale optimization algorithm (IWOA) enhances the efficiency by balancing the exploration and exploitation capabilities of WOA.
The proposed IWOA method are used for optimizing the hyper-parameters of neural network and to identify the stance of textual data.
The remaining paper is ordered as follows. The section “Background study” briefs the related work in the field of stance detection. Preliminaries are discussed in the section “Preliminaries” and the proposed work is presented in the section “Proposed algorithms”. The section “Evaluating IWOA for bias(es)” evaluates for the bias(es) and the section “Experimental results” reports the Experimental outcomes followed by the conclusion in the section “Conclusion and future work”.
Anzeige
Background study
Recently, many deep neural network-based techniques have been proposed for stance detection. Riedel et al. [68] proposed a two-step model for stance detection. The authors first extracted and created a vocabulary set of 5000 most frequent words using TF-IDF which are then passed to a multilayer perceptron having a single hidden layer to get the stance. The proposed technique shows acceptable results for agreeing label stance. However, it does not give satisfactory results for other stances. Therefore, Davis and Proctor [21] presented three different approaches for correctly identifying the stances of the news; the first approach is based on bag-of-words with a three-layer multilayer perceptron (BoW-MLP), the second is a bag of words with bidirectional LSTM (BoW-BiLSTM), and the third is a bag of words with a concatenated multilayer perceptron (BoW-CMLP). The proposed BoW-MLP model outperforms the other two proposed BoW models with a classification accuracy of 93%. However, the proposed models do not perform well with information captured by the BoW model. Thus, Chen et al. [14] used a vectorization technique based on the strength of the LSTM and attention model to predict stances. For the same, they used three different types of neural network architecture with a bag of vectors. The results show that the bag of vector technique performs well with neural network for related and agree news articles. However, it does not perform well in classifying the stance of discuss and unrelated category.
Furthermore, Chaudhry et al. [12] used GloVe word embedding along with LSTMs and three different encoding schemes, namely conditional, independent, and bidirectional conditional for stance detection. In LSTM with conditional encoding, RNN is arranged sequentially, while in LSTM with bidirectional conditional encoding, the softmax layer is used for prediction. The results show that LSTM with bidirectional conditional encoding outperforms the other two algorithms with an accuracy of 97%. However, the accuracy can be further improved by combining an attention mechanism with high-dimensional pre-trained embedding. Mrowca et al. [52] used 100-dimensional GloVe vector representations with bidirectional LSTM for stance detection. Pfohl et al. [65] used the notions of attention model and conditional encoding-based LSTM for identifying the stance of news articles. Furthermore, basic LSTM, LSTM with attention, and conditional encoding LSTM with attention (CEA-LSTM) have been used for stance detection [9, 58]. Zeng et al. [86] compared the performance of deep neural network models with a hand-crafted feature-based system and discovered that the deep neural network model outperformed the hand-crafted feature-based system. Sun et al. [74] used a hierarchical attention network for identifying the stance of news. The performance of the attention model is further improved by combining the hand-crafted features and hidden features. Additionally, Yu et al. [85] implemented an RNN encoder–decoder using LSTM and GRU cells with and without having attention for stance detection. In the proposed model, LSTM and GRU cells allow the flexibility to remember or forget the context and also deal with the vanishing gradient problem. Yoon et al. [84] detected incongruity between the body and headline of news articles using a deep hierarchical encoder.
Moreover, Le and Mikolov [42] combined doc2vec with Word2vec word embedding for stance detection. Besides, Lau and Baldwin [41] presented an improved bag-of-words model for stance detection. However, the existing machine learning approaches use some predefined lexicons and word embeddings to extract features from textual data. Thus, it is possible that the extracted features may be irrelevant wherever deep learning models suffer from vanishing gradient descent problem. Therefore, to overcome the problem of vanishing gradient descent, metaheuristic algorithms are generally used [7, 24, 57]. Metaheuristic methods normally show better results than the traditional and state-of-the-art methods over NP problems [11, 57, 59‐61]. Mosavi et al. [51] introduced a hybrid model based on gray wolf optimization and neural network for the classification of sonar data. Mukherjee et al. [53] introduced a PSO optimized multilayer perceptron for malignant melanoma detection. Kohli and Arora [38] presented a chaotic GWO algorithm to enhance the global convergence speed of GWO. Krill herd algorithm has been also used to solve a number of optimization problems. Abualigah [2] employed an improved krill herd algorithm for feature selection and document clustering. Cuevas et al. [20] proposed a social spider algorithm based on the behavior of spider to solve many real-world optimization problems like economic dispatch problem [23], transmission expansion planning problem [22], and optimal power flow solution with single-objective optimization [54]. Furthermore, a new metaheuristic algorithm namely symbiotic organisms search (SOS) has been proposed to solve engineering design and various numerical optimization problems [17]. SOS algorithm has been used to solve complex real-world optimization problems such as predicting sea wave height [3], truss optimization with natural frequency [75], optimal operation of reservoir systems [10], and many more. In continuation, Mirjalili et al. [46] proposed a new bio-inspired optimization algorithm, namely salp swarm optimization algorithm (SSA) for engineering design problems. A number of variants of SSA have presented in the literature for solving various NP problems. Yılmaz et al. [82] presented a biobjective optimization model that minimizes the makespan and reduces the workload imbalance among workers for seru production system. Yilmaz and Durmusoglu [83] compared the performance of different metaheuristics for batch scheduling problem in a multihybrid cell manufacturing system. The distribution optimization approaches have also been used for finding optimal parameters for numerous applications. Sun et al. [73] presented a wind forecasting approach based on two-step short-term probabilistic distribution optimization. WOA is also employed in a number of NP problems. WOA has been proposed by Mirjalili and Lewis [44], which is motivated by the hunting mechanism of humpback whales. Mafarja and Mirjalili [43] proposed a feature selection algorithm based on WOA and simulated annealing. Alzaqebah et al. [6] employed WOA to prioritize the software requirements by assuming requirements as a search space and priority as hunting behavior of the whales. Petrović et al. [64] presented a new variant of WOA to find the optimal solution for the NP-hard scheduling problem. Jiang et al. [37] introduced a discrete WOA to solve the green job shop scheduling problem. A new binary WOA based on S-shaped and V-shaped transfer functions has been proposed by Hussien et al. [35] to solve discrete optimization problems. Chen et al. [15] employed levy flight and chaotic local search synchronously to balance the exploration and exploitation capabilities of standard WOA. The proposed balance WOA has been used to solve complex constrained engineering design problems. Aljarah et al. [5] used a whale optimization algorithm (WOA) along with an MLP for identifying the stance of news. For the same, the authors employed the WOA algorithm for updating the weights and biases for backpropagation instead of gradient descent. The balance WOA proposed by Aljarah et al. [5] for optimizing the MLP mitigates the issue of vanishing gradient problem and also enhances the convergence speed. From the literature, it has been observed that WOA has been successfully used to solve diverse real-word optimization problems including single-objective and multi-objective problems [1, 16], especially in text-mining tasks [4, 36, 49, 78]. Therefore, in this study, WOA has been used for identifying the stance of fake news.
From the above discussion, it is evident that the traditional machine learning and deep learning models generally suffer from vanishing gradient descent problems and computationally very expensive. Moreover, these methods also perform badly over NP problems. On the contrary, metaheuristic methods do not suffer from gradient descent problem and also show better results on NP problems. In literature, a number of metaheuristic methods have been introduced for optimizing the different types of neural networks and for solving the various real-world problems. Therefore, in this paper, a new variant of WOA named IWOA has been proposed. The proposed IWOA can be used for solving various engineering design and NP problems like truss optimization, a traveling salesman, cell manufacturing system, text classification, and many more. In this work, the proposed IWOA optimized has been used for optimizing a neural network that has been utilized for stance detection for fake news.
Anzeige
Preliminaries
The proposed approach uses an improved whale optimization algorithm (IWOA) for optimizing the weight and biases of multilayer perceptron which are discussed in the following subsections.
Whale optimization algorithm
Whale optimization algorithm (WOA) is a nature-inspired metaheuristic algorithm which is generally used for solving different optimization problems [44]. WOA is motivated by the hunting mechanism of humpback whales. WOA method uses three phases, i.e., encircling the prey, bubble-net attacking, and searching of prey to find the optimal solution. All the phases of WOA are discussed below.
Encircling the prey
In this phase, each of the search agents update their positions vector concerning the position vector of current best agent using Eqs. (1) and (2):
where t denotes the iteration, \(\mathbf {A}\) and \(\mathbf {B}\) are coefficient vectors, \(\mathbf {W^*}\) is the position vector of the best solution obtained so for, and \(\mathbf {W}\) denotes the position vector of a whale. The values of \(\mathbf {A}\) and \(\mathbf {B}\) are computed according to Eqs. (3) and (4):
where \(\mathbf {a}\) is a control parameter and \(\mathbf {d}\) is a random vector in the range of [0, 1].
Bubble-net attacking
The attacking mechanism performed by a humpback whale is also known as a bubble-net attack. It is basically an exploitation phase in which the obtained solution is further refined for finding the optimal solution. The bubble-net attacking has two phases, i.e., shrinking encircling medium and spiral updating position mechanism. For the course of the next iteration, one of these mechanisms is chosen based on the probability. For the same, the first probability is randomly generated, and if the probability is less than 0.5, then the shrinking encircling mechanism is used; otherwise, the spiral updating position mechanism is used. Both shrinking encircling and spiral updating position mechanisms are discussed in the following paragraph.
Shrinking encircling mechanism To achieve the shrinking encircling behavior, the value of \(\mathbf {a}\), as defined in Eq. (3), is decreased linearly from 2 to 0. In other words, A is set to \([-a, a],\) where \(\mathbf {a}\) is decreased from 2 to 0 over the course of iterations. Therefore, the new position of a search individual using this approach can be defined anywhere in between the current best individual and the original position of the search individual.
Fig. 1
Block diagram of MLP
×
Spiral updating position mechanism In this phase, the search agents move in helix shape toward the prey using Eqs. (5) and (6):
here, \(W^*\) is position vector of the best search agent, \(\mathbf {W(t)}\) is position vector of search agent in iteration t, \(\mathbf {Q'}\) is coefficient vector which is computed using Eq. (5), b is a constant value, and k is a random number in range \([-1,1].\)
Search for prey (exploration phase)
In this phase, a search agent searches for prey randomly. This random movement serves two purposes; first, it helps in proper investigation of the search region, and the other, it helps to avoid algorithm form being stuck at local optima. Mathematically, it can be formulated using Eqs. (7) and (8):
where \(\mathbf {W}_{\mathrm{rand}}\) is random vector, \(\mathbf {W(t)}\) is position vector of search agent at iteration t, \(\mathbf {A}\) and \(\mathbf {B}\) are position vectors, and \(\mathbf {Q}\) is coefficient vector.
Multilayer perceptron neural network
A multilayer perceptron (MLP) is a neural network that consists of multiple perceptrons as depicted in Fig. 1. MLP consists of an input layer to accept the inputs, an output layer that predicts the input, and hidden layers [55]. The hidden layer is the true computing engine of the MLP which learns more complicated features of data. MLP is generally used for supervised learning problems in which they are trained on a variety of input–output pairs to learn the dependencies (or correlation) between those inputs and outputs. Feedforward neural networks (FFNNs) are a specific form of MLP [25]. In FFNNs, neurons are interlinked in a one-way and one-directional manner. Connections are depicted by weights that are real numbers and fall in the range \([-1, 1].\) Figure 1 illustrates an FFNN with only one hidden layer.
Proposed algorithms
In this paper, weights and hyper-parameters of MLP have been optimized using an improved whale optimization algorithm for enhancing the efficiency of stance detection. The next subsections describe the proposed improved whale optimization algorithm followed by the proposed stance detection method.
Improved whale optimization algorithm
The success of metaheuristic methods depends upon diversification and intensification steps [48, 56, 61, 62]. The metaheuristic methods which maintain the trade-off between diversification and intensification are considered superior. In whale optimization algorithm (WOA), search individual (humpback whales) that guides the search process is randomly selected. Therefore, WOA usually suffers from slow convergence. Moreover, it also sticks at some local solutions due to hastening exploitation. Thus, in this paper, a new variant of WOA named improved whale optimization algorithm (IWOA) has been introduced to boost the convergence speed and to attain better results. The proposed variant is invigorated with the capabilities of WOA, roulette wheel, and tournament selection. The proposed improved whale optimization algorithm is discussed below.
Improved search for prey using hybrid TS–RS selection
In the intensification (exploitation) phase of WOA, humpback whales that control the entire search process are randomly selected. However, due to this random selection of humpback whales, WOA sometimes takes a longer time to find the optimal solution. Therefore, to equalize the exploration and exploitation and boost the convergence speed, the proposed IWOA uses tournament selection and roulette wheel selection [43] in alternate iterations. In tournament selection, search agents with the best fitness are selected from a group and these search agents guide the whole search process. However, the tournament selection-based WOA sometimes traps in a local solution if local search agents are selected as humpback whales. Thus, to maintain the diversity, roulette wheel selection [49] and tournament selection are used alternatingly to select the humpback whales. The proposed IWOA uses the following steps to find the searched individual that guides the whole search process. First, tournament selection is used in odd iteration to find the search agents (humpback whales). Tournament selection employs the following steps to select the searched individual:
1.
Select few search individuals randomly from the population (a tournament).
2.
Compute fitness of each search individual and sort them according to their fitness.
3.
The search individual with the best fitness value (the winner) is selected.
Secondly, roulette wheel selection is used to control the trade-off between exploration and exploitation. To understand mathematically, consider there are n search agents in a population \(P_s= \{s_1, s_2, \ldots , s_n\}\) and fitness value of each search agents \(s_i\) is \(f(s_i)\). First, tournament selection is used to elect the humpback whales. Second, roulette wheel selection is employed in which selection probability (\(P_m(s_i)\)) as well as cumulative probability \((P_l{(s_k)})\) as given in Eqs. (9) and (10) are used to select the search agents:
here, \(\mathbf {\mathrm{W}}_{{\mathrm{RS}}}\) and \(\mathbf {\mathrm{W}}_{{\mathrm{TS}}}\) is elected by the roulette wheel and tournament selection, respectively. The improved WOA is depicted in Algorithm 1. Furthermore, the proposed IWOA has been used to optimize the neural network for enhancing the efficiency of stance detection.
Fig. 2
Structure of three-layer FNNs
×
×
Proposed stance detection method
Nowadays, metaheuristic methods are also used for training multilayer perceptron neural networks (MLPNN) to improve their performance. Metaheuristic-based neural network differs from traditional multilayer perceptron as the weight and biases are updated by the metaheuristic algorithms instead of gradient descent algorithms. Hence, these methods generally do not suffer from vanishing gradient and exploding gradient problems in large neural networks. Metaheuristic methods are applied to three aspects of neural networks. First, metaheuristic methods are applied to find the combination of weights and biases that minimizes the error. Second, metaheuristic methods find a proper structure of FFNN for a problem. Finally, parameters such as momentum and learning rate are tuned for gradient-based learning. In this work, an improved whale optimization algorithm-based stance detection method (IWOASD) has been introduced for finding the optimal values of weights and biases. The proposed IWOASD is based on the capabilities of IWOA and MLP. The most important factor in the proposed IWOASA is to find the optimal values for weights and biases. To achieve the objective, the input values given to the IWOA should be in the form of vector as given in Eq. (14):
where m is the number of the input nodes, \(W_{ij}\) is the connection weight from the node i to j, and \(b_{j}\) is the bias.
The next step is to define the objective function of the proposed IWOASD algorithm. The objective function for the proposed IWOASD has been discussed below.
Objective function Consider an FFNN depicted in Fig. 2 with three layers (one hidden, one input, and one output) has been given. The output of each hidden in every epoch is computed according to Eq. (15):
here, \(W_{kj}\) is the connection weight from kth output node to jth hidden node.
Finally, mean square error (MSE) is computed using Eq. (17). The value of MSE is the difference between MLP output \(O^k\) and desired output \(d^k\). Furthermore, for making the FFNN more effective, the efficacy of the FFNN is computed by taking the average value of MSE over every training samples using Eq. (18):
where S is the number of training instances, \(d^k_{j}\) is predicted output, and \(O^k_{j}\) is the actual output of jth input when kth training sample is used.
To find the combination of optimal weights and biases, the proposed IWOASD algorithm uses MSE defined in Eq. (18). To design an optimized neural network, the encoding strategy also needs to be defined.
Encoding strategy In the literature, there are three encoding and representing methods, namely vector, matrix, and binary for representing weights and biases in metaheuristic methods. In a vector scheme, every agent is encoded as a vector, while in a matrix scheme, every search individual is encoded as a matrix. On the other hand, in the binary encoding scheme, each search individual is represented as strings of binary bits. The decoding particle’s vectors of biases and weights in the vector scheme is a complicated task. On the contrary, the binary encoding strategy represents particle variables in binary form in which the length of particle grows for the complex neural network structure, while the decoding strategy of the matrix approach is simple and easy to execute. Therefore, in this work, matrix encoding strategy has been used.
The detailed steps of stance detection using an optimized neural network are presented in Algorithm 2.
×
Fig. 3
Flow graph for optimized neural network
×
The workflow of the proposed model can be summarized as follows.
1.
First, weights of MLPNN are generated randomly using IWOA
2.
Second, fitness of all the individuals are computed using the objective function that minimizes the mean squared error as given in Eq. (18)
3.
Weights are updated using IWOA
4.
Steps 2–3 are repeated till convergence.
The entire steps of the proposed IWOASD algorithm are depicted in Fig. 3. From the figure and above discussions, it has been observed that the proposed IWOASD algorithm optimizes the weight and biases of MLP by minimizing the MSEs. Considering MLP part similar for all the algorithms, the computational complexity of the IWOA part can be defined as \(O(I \times D\times N^2)\), where I is the current iteration, D is the dimension, and N is the total number of individuals (whales) in population. The computational complexity of FFNN depends upon training and an inference phase. The total time taken to train (backpropagation) an FFNN with n nodes will be \(O(n^5)\) and \(O(n^4)\) for forward propagation or inference phase. Thus, it can be perceived that the backpropagation phase is much slower than forward propagation. Due to this issue, nowadays, pre-trained neural networks are generally used.
Evaluating IWOA for bias(es)
Generally, the metaheuristic method shows biased results on a number of benchmark functions. Thus, it is better to get an idea of the inherent bias(es) of metaheuristic methods before testing their performance on benchmark functions. Metaheuristic methods may be central bias, edge bias, or/and axial bias. In central bias, solutions closed to the center are explored, while in edge bias, solutions near the boundary (edges) are explored. On the other hand, axial bias search solutions along the x- and y-axis. Therefore, signature test [19] has also been carried out to inspect the bias(es) of the proposed IWOA and standard WOA. For the same, the minimization problem is given in Eq. (19) is used in which every point in the search region represents an optimal solution and an unbiased metaheuristic method produces solutions similar to random search. The outcome of the signature test for WOA and IWOA is depicted in Fig. 4. From the figure, it can be envisioned that the IWOA explores the entire search space evenly and unbiased, while WOA shows center biased results:
The performance of the proposed stance detection method is discussed in the following subsections. First, the section “Performance analysis of IWOA” investigates the efficacy of the IWOA over unimodal and multimodal benchmark functions. The section “Performance estimation of stance detection method” discusses the performance analysis of IWOA-optimized neural network on five stance detection benchmark datasets. All the experiments have been performed on Matlab 2017a on a computer with 8 GB of RAM and a 2.66 GHz core i3 processor.
Performance analysis of IWOA
The efficiency of IWOA has been evaluated over 17 benchmark functions including both unimodal (\(F_1\)–\(F_{10}\)) and multimodal (\(F_{11}\)–\(F_{17}\)) functions [60, 72]. Unimodal functions generally appraise the convergence performance, while multimodal functions investigate the probability of trapping into local solution. The considered unimodal and multimodal benchmark functions are depicted in Table 1. To assess the efficiency of the proposed IWOA, mean fitness values of IWOA and other state-of-the-art metaheuristic algorithms, namely cuckoo search (CS) [40, 80], gray wolf optimization (GWO) [45], whale optimization algorithm (WOA) [44], grasshopper optimization algorithm (GOA) [47], bat algorithm (BA) [81], hybrid cuckoo search (CSK) [61], vector co-evolving particle swarm optimization (VCPSO) [87], global and local neighborhood-based PSO (GLNPSO) [18], scout particle swarm optimization (ScPSO) [39], random spare reinforced whale optimization algorithm (RDWOA) [16], hybrid whale optimization algorithm (HWOA) [1], naive Bayes-based whale optimization algorithm (NB-WOA) [34], and enhanced whale optimization algorithm (EWOA) [49] have been computed. For a fair comparison, the controlling parameters such as population dimension (\(P_{{\mathrm{dims}}}\)) is set to 30 and 50 and max iteration (mitr) is 1000, for all the algorithms. The other parameters are taken from their respective literature and the values of parameters for the proposed algorithm have been decided empirically by testing its performance on different parameter values of standard WOA algorithm.
As nature-inspired algorithms are randomized by their behavior, therefore all the algorithms have been run 30 times and their average values are used for comparison. The average value of fitness of all the algorithm is displayed in Tables 2 and 3. It is observed from the tables that the proposed IWOA demonstrates the best outcomes than the state-of-the-art methods on 85% of the benchmark functions. There are a few functions for which other algorithms show better results than IWOA. EWOA shows comparable performance on \(F_1\), \(F_{13}\), and \(F_{15}\) for only one dimension and on both dimensions of function \(F_3\), while WOA performs better than the proposed IWOA for dimensions 30 and 50 over \(F_7\) and \(F_{14}\) benchmark functions, respectively. RDWOA achieves the best fitness value for the benchmark functions \(F_{5}\) and \(F_{6}\), respectively, for dimension 50, while for the benchmark functions \(F_{10}\) and \(F_{11}\), ScPSO, RDWOA, HWOA, EWOA, and NB-WOA show equivalent performance. Hence, it is evident that the proposed IWOA performs better than the compared methods.
Table 2
Comparative analysis of existing and proposed algorithms for mean fitness over 30 runs on the benchmark functions (\(F_1\)–\(F_{9}\)) for 30 and 50 dimensions
Funs.
Dims.
CS
GWO
BA
WOA
GOA
CSK
VCPSO
GLNPSO
ScPSO
RDWOA
HWOA
NB-WOA
EWOA
IWOA
\(F_1\)
30
4.09E−01
5.98E−01
5.06E−03
1.06E−02
1.64E−01
8.64E−04
4.32E−03
2.22E−03
1.44E−03
1.72E−04
7.98E−04
2.81E−04
8.91E−05
2.97E−06
50
2.87E\(+\)00
8.11E\(+\)00
9.50E−03
9.73E−04
8.77E−01
4.66E−02
2.34E−01
7.29E−04
2.27E−03
4.59E−04
2.17E−04
1.59E−04
1.23E−04
2.63E−03
\(F_2\)
30
8.99E−06
1.09E−05
8.03E−06
2.83E−05
1.96E−06
3.54E−07
1.77E−06
3.71E−06
1.70E−07
1.24E−07
3.05E−06
3.40E−07
4.83E−08
1.16E−08
50
2.20E−05
3.69E−05
3.27E−05
5.90E−08
6.58E−06
8.49E−07
5.10E−06
3.29E−05
1.89E−06
1.55E−06
2.08E−05
6.10E−06
9.44E−07
2.68E−08
\(F_3\)
30
3.74E−08
7.04E−08
8.47E−08
6.25E−08
3.20E−08
2.00E−08
1.00E−07
3.85E−08
1.39E−09
1.79E−08
2.38E−08
6.72E−10
9.23E−11
2.82E−08
50
4.11E−08
7.78E−08
8.78E−08
3.94E−07
3.49E−08
1.98E−08
1.19E−07
2.23E−08
5.05E−09
1.35E−09
1.64E−08
2.21E−09
8.40E−12
2.82E−09
\(F_4\)
30
2.76E−07
3.24E−07
4.42E−07
1.14E−06
2.14E−07
3.03E−08
1.82E−07
4.09E−07
5.37E−08
3.22E−08
3.69E−07
3.67E−07
2.30E−08
1.87E−09
50
9.54E−07
1.60E−06
1.20E−06
9.64E−08
6.92E−07
5.42E−08
3.26E−07
6.80E−07
2.56E−08
5.21E−09
9.63E−08
8.31E−08
7.48E−10
4.25E−10
\(F_5\)
30
6.04E−08
1.14E−07
1.38E−07
9.64E−08
5.19E−08
1.21E−08
1.64E−07
6.27E−08
5.98E−08
5.53E−08
2.41E−08
2.39E−08
7.22E−09
3.27E−10
50
6.06E−08
1.14E−07
1.38E−07
1.14E−07
5.20E−08
4.69E−09
1.96E−07
1.11E−07
3.69E−09
7.02E−10
8.77E−08
6.13E−08
9.67E−07
1.27E−08
\(F_6\)
30
7.24E−08
1.36E−07
1.67E−07
1.14E−07
6.23E−08
3.95E−08
2.38E−07
1.34E−08
4.20E−08
1.70E−08
9.88E−09
6.47E−10
1.30E−08
9.06E−09
50
7.25E−08
1.37E−07
1.67E−07
1.33E−07
6.23E−08
3.94E−08
2.37E−07
1.10E−07
4.58E−09
2.76E−09
7.48E−08
6.63E−08
6.33E−08
5.67E−06
\(F_7\)
30
8.71E−08
1.63E−07
1.94E−07
1.33E−10
7.41E−08
4.80E−08
2.41E−07
9.88E−08
4.32E−08
3.56E−08
8.73E−08
1.27E−08
1.78E−09
6.59E−08
50
9.06E−08
1.78E−07
1.94E−07
1.59E−07
7.46E−08
7.86E−08
3.93E−07
1.83E−07
5.46E−08
1.12E−08
5.45E−08
4.32E−08
5.19E−08
6.78E−09
\(F_8\)
30
4.99E−07
2.65E−07
2.23E−07
2.22E−07
1.69E−07
1.38E−07
6.88E−07
1.66E−07
1.52E−08
1.60E−07
6.53E−07
2.28E−07
4.58E−07
2.00E−07
50
9.60E−05
1.67E−03
2.45E−07
1.74E−07
1.31E−05
9.20E−06
4.60E−05
1.30E−07
1.87E−05
1.19E−05
4.10E−08
7.38E−09
2.07E−06
1.85E−09
\(F_9\)
30
6.93E−13
3.72E\(+\)01
1.66E−17
1.32E−20
6.48E\(+\)01
1.21E\(+\)01
7.64E−18
5.46E−12
9.07E−33
8.43E−33
1.63E−34
1.01E−35
2.92E−33
1.84E−43
50
3.46E−13
1.86E\(+\)01
5.53E−18
4.40E−21
3.24E\(+\)01
4.05E\(+\)00
3.39E−18
1.81E−12
1.52E−33
2.71E−34
1.21E−33
6.68E−32
7.50E−33
1.12E−34
Best values are present in bold
Table 3
Comparative analysis of existing and proposed algorithms for mean fitness over 30 runs on the benchmark functions (\(F_{10}{-} F_{17}\)) for 30 and 50 dimensions
Funs.
Dims.
CS
GWO
BA
WOA
GOA
CSK
VCPSO
GLNPSO
ScPSO
RDWOA
HWOA
NB-WOA
EWOA
IWOA
\(F_{10}\)
30
0
0
0
2.56E−75
0
1.70E−75
2.27E−76
5.12E−76
0
0
0
0
0
0
50
9.98E−79
3.84E−85
0
0
0
0
0
0
0
0
0
0
0
0
\(F_{11}\)
30
7.64E−05
3.68E\(+\)01
1.52E−05
0
2.15E−06
1.30E−04
9.57E−76
7.94E−05
0
0
0
0
0
0
50
8.58E−03
4.38E\(+\)03
1.58E−03
2.27E−14
6.80E−04
1.45E−02
0
6.78E−03
0
0
0
0
0
0
\(F_{12}\)
30
3.89E−15
1.59E−04
1.67E−13
2.34E−16
1.19E−13
1.31E−14
1.00E−30
1.24E−13
1.05E−30
9.01E−31
2.60E−31
2.34E−31
2.58E−33
9.33E−43
50
9.72E−16
2.65E−05
5.55E−14
2.93E−17
2.29E−14
3.29E−15
2.66E−32
2.49E−14
3.27E−32
1.72E−32
8.13E−33
2.11E−32
4.05E−33
1.16E−34
\(F_{13}\)
30
9.29E−08
1.53E−07
2.15E−07
1.50E−07
2.51E−07
6.30E−08
1.54E−08
8.50E−08
1.86E−08
6.01E−09
3.66E−09
4.26E−09
1.91E−12
8.21E−09
50
9.46E−08
1.54E−07
2.18E−07
1.45E−04
3.05E−07
1.47E−07
1.55E−08
1.08E−07
2.61E−08
1.71E−08
8.06E−09
1.57E−09
5.67E−09
8.28E−10
\(F_{14}\)
30
1.40E−04
2.49E−04
2.06E−05
1.22E−03
1.46E−05
1.06E−05
7.48E−07
1.02E−05
7.52E−06
6.78E−06
4.73E−06
4.48E−07
4.79E−09
1.79E−10
50
4.00E−03
1.61E−02
4.31E−04
1.83E−10
8.75E−06
2.72E−04
5.35E−07
1.71E−04
1.63E−04
7.32E−06
4.01E−06
0
1.72E−08
2.92E−05
\(F_{15}\)
30
1.15E−07
1.89E−07
2.62E−07
1.86E−07
3.10E−07
1.81E−09
1.90E−08
1.04E−07
3.10E−08
5.91E−09
5.66E−09
3.29E−09
3.61E−09
9.50E−10
50
1.33E−07
2.42E−07
2.68E−07
2.45E−07
3.90E−07
1.29E−07
1.33E−08
1.06E−07
9.45E−09
5.75E−09
4.06E−09
6.13E−09
1.33E−09
1.04E−08
\(F_{16}\)
30
1.65E−07
2.67E−07
3.26E−07
3.36E−07
4.72E−07
3.87E−08
2.41E−08
1.29E−07
7.31E−08
3.48E−08
1.29E−08
9.87E−09
1.16E−09
1.10E−10
50
3.11E−07
6.45E−07
4.11E−07
2.71E−07
4.98E−07
1.25E−07
2.55E−08
1.62E−07
4.84E−08
2.40E−08
2.70E−09
1.15E−09
6.70E−09
1.16E−10
\(F_{17}\)
30
1.81E\(+\)01
1.62E\(+\)02
4.69E−09
1.43E−10
2.70E−09
1.07E\(+\)02
3.70E−10
9.94E\(+\)01
2.98E−09
1.91E−09
1.47E−09
3.28E−11
2.97E−11
5.37E−12
50
1.29E\(+\)00
2.31E\(+\)01
9.37E−10
3.58E−11
3.41E−10
5.47E\(+\)00
4.35E−11
2.90E\(+\)00
5.32E−11
4.71E−11
1.60E−11
1.48E−11
1.54E−12
1.34E−12
Best values are present in bold
Table 4
Mean ranking of all the considered methods
Rank
Methods
Rank value
1.
IWOA
1.84
2.
EWOA
1.98
3.
RDWOA
2.07
4.
NB-WOA
2.15
5.
ScPSO
2.38
6.
HWOA
2.47
7.
VCPSO
2.72
8.
GLNPSO
2.96
9.
GWO
3.18
10.
CSK
3.37
11.
BA
3.68
12.
WOA
3.83
13.
GOA
4.08
14.
CS
4.25
Best values are present in bold
Furthermore, the non-parametric Friedman’s test [89] has also been conducted in Table 4 to statistically validate the efficacy of IWOA. Friedman’s test assesses the efficiency of each method on every benchmark function and ranks them according to their performance [79]. The best performing method gets the rank of 1, the next best obtain rank 2, then \(3,\ldots ,n\). If two methods have the same performance, then the rank is computed by averaging the ranks returned in different runs [70]. The p value returned by the Friedman test is 0.0004518, which is much smaller than the threshold \((\alpha = 0.05)\) which signifies that the obtained results are significantly different. Table 4 tabulates the ranks of all the methods returned by the Friedman test. From Table 4, it is evident that the proposed IWOA has a minimum ranking value among all the considered models. Hence, the experimental and statistical outcomes reflect the high efficiency of IWOA.
Performance estimation of stance detection method
The performance of the proposed IWOASD method has been analyzed on five stance datasets namely, argument reasoning comprehension (ARC) dataset [31], headline and article bodies dataset (FNC-1) [52], claim polarity dataset Research [67], perspectrum dataset [69], and Snopes corpus [30]. The ARC dataset is manually created by Habernal et al. [29] and it consists of 188 debate topics from the user debate section of the New York Times, while the FNC-1 dataset is obtained from Fake News Challenge (FNC-1). Both ARC and FNC-1 datasets are annotated in four categories, namely, unrelated, discuss, agree, and disagree. On the other hand, claim polarity, perspectrum, and snopes corpus are annotated in two classes, viz, support (agree) and contest (disagree). All the datasets consist of pairs of claims and evidence texts along with stances. Stance in data may belong to any one of the categories, i.e., unrelated, discuss, agree, or disagree. The complete description of datasets is given in the following subsections.
FNC-1 dataset
This dataset is motivated by an Emergent dataset that originated from a digital journalism project at Columbia University [26]. The emergent dataset comprises of 2595 news articles and 300 rumored claims and grouped into true, false, and unverified categories by journalists. The FNC-1 dataset [77] expands the Emergent dataset by allocating four labels to each headline–body pair, namely, unrelated, discuss, agree, and disagree.
Table 5
The dataset statistics
Dataset
Data types
Attribute types
Number of samples
Unrelated (%)
Discuss (%)
Agree (%)
Disagree (%)
Date and time
Associated tasks
FNC-1 dataset
Multivariate, text
Real
49,972
73.13
17.83
7.36
1.68
2017-06-22
Classification
ARC dataset
Multivariate, text
Real
17,792
75.0
6.1
8.9
10.0
2017-06-22
Classification
Claim polarity dataset
Multivariate, text
Real
2394
–
–
55.30
44.70
2020-08-01
Classification
Perspectrum dataset
Multivariate, text
Real
11,876
–
–
51.60
48.40
2020-08-01
Classification
Snopes dataset
Multivariate, text
Real
8291
–
–
74.50
24.50
2020-08-01
Classification
ARC dataset
This dataset contains typical disputed subjects namely, schooling challenges, immigration, and international affairs from different news domains. This dataset is equivalent to the FNC-1 dataset with some considerable differences. In the FNC-1 dataset, the news articles are more balanced and complete, while in the ARC dataset, multisentence statement represents users’ perspective of a topic. Table 5 tabulates the complete statistics of FNC-1 and ARC datasets. It can be visualized from the table that both the datasets are skewed. In the literature, it has been shown that the poor models show unsatisfactory results on skewed datasets [13, 57]. Therefore, the efficacy of the proposed IWOA has been evaluated on skewed (imbalanced) datasets to show its performance.
Claim polarity dataset
This dataset contains 2394 evidences and claims for 55 topics Research [67]. Topics in the dataset were randomly selected from the debate motions database [8]. In the dataset, all the claims are manually annotated in two classes, namely, support (agree) and contest (disagree). This dataset comprises 1324 support claims and 1070 contest claims.
Perspectrum dataset
This dataset contains the users’ views and perspective from various debate websites such as debatewise.org, idebate.com, and procon.org [66]. Each claim in the dataset has different views and stances. The dataset consists of 6125 supporting (agree) claims and 5751 opposing (disagree) claims [66, 69]. For a fair comparison, the dataset is divided into three parts, i.e., training data, dev data, and test data. The complete dataset statistics have been tabulated in Table 5.
Snopes dataset
This dataset has been collected from the Snopes platform and consists of 8291 claim [30, 33]. The dataset is annotated into two classes, namely, agree and refutes. An instance in the dataset is annotated as agree (support) if the claim is supported by evidence text; otherwise, it is annotated as refute or disagree. There is a total of 6178 agree instances and 2113 refute instances in the dataset [33].
To assess the efficiency of the proposed IWOASD, the first datasets are partitioned into training and testing parts in which each instance of the dataset contains a pair of claim and evidence texts. Since stop words, fuzzy words do not contain relevant information. Therefore, the stance datasets are preprocessed for eliminating unwanted words such as ‘a’, ‘is’, ’are’, etc. from the datasets. Afterward, vocabulary lists are built for distinct terms occurring in headlines and article texts which are further used for extracting pre-trained word embeddings from the Stanford Glove corpus. The pre-trained word embeddings obtained from training sets are used for training the proposed IWOASD method. Finally, the test dataset is fed to a trained model to examine the efficiency of the proposed IWOA-based MLP.
The efficacy of the proposed IWOASD model is evaluated in respect of mean classification accuracy, mean error, standard deviation, and mean computational time for different numbers of hidden nodes (\(5, 7,9,\ldots ,20\)), and compared with CS, GWO, BA, WOA, and CSK. Mean classification accuracy is computed using a confusion matrix. In the confusion matrix \(C_x\) of size \(m \times m\), \(C_{kk}\) represents the number of samples of class k predicted to same class, i.e., k. In the confusion matrix, diagonal entries represent correctly predicted entries. The mean accuracy, mean error, standard deviation values, and mean computational time of the proposed and state-of-the-art methods have been tabulated in Tables 6, 7, 8, 9, and 10. From the tables, it can be observed that the proposed IWOA with 20 hidden nodes returns the highest accuracy on the FNC-1 dataset and snopes corpus, while the proposed IWOA with 11 hidden nodes attains the highest accuracy over the ARC, claim polarity, and perspectrum datasets. The proposed IWOA also surpasses other methods in terms of mean error, while for the performance measure standard deviation, other methods such as CS, BA, and CSK show better results than the proposed method. Moreover, the proposed IWOA also performs better than the considered methods for performance criterion and computational time.
Table 6
Comparison of mean accuracy, standard deviation, mean error, and execution time over FNC-1 dataset for the different numbers of hidden nodes
Hidden node(s)
Algorithm
Mean accuracy
Standard deviation
Mean error
Execution time
5
CS
69.17
2.04
0.3484
15 min 58 s
GWO
70.54
2.32
0.3046
17 min 51 s
BA
69.54
2.06
0.3237
18 min 27 s
WOA
70.54
2.96
0.3360
19 min 02 s
CSK
71.54
2.76
0.2931
18 min 25 s
Proposed IWOA
75.97
3.06
0.2408
13min03s
7
CS
68.75
2.34
0.3460
26 min 22 s
GWO
69.71
2.32
0.3022
28 min 10 s
BA
69.13
2.04
0.3213
28 min 43 s
WOA
69.71
2.96
0.3336
29 min 19 s
CSK
71.13
2.76
0.2907
28 min 56 s
Proposed IWOA
76.04
3.06
0.2384
23min27s
9
CS
69.22
2.32
0.3463
36 min 45 s
GWO
70.67
2.29
0.3025
38 min 37 s
BA
69.87
2.02
0.3216
38 min 51 s
WOA
70.80
2.92
0.3339
39 min 43 s
CSK
71.52
2.73
0.2910
39 min 06 s
Proposed IWOA
75.83
3.03
0.2387
33min32s
11
CS
68.99
2.33
0.3462
56 min 34 s
GWO
70.19
2.31
0.3024
58 min 24 s
BA
69.50
2.03
0.3215
58 min 47 s
WOA
70.26
2.94
0.3338
59 min 31 s
CSK
71.32
2.14
0.2909
59 min 31 s
Proposed IWOA
75.94
3.04
0.2385
53min30s
13
CS
69.07
2.33
0.3472
66 min 45 s
GWO
70.28
2.30
0.3034
68 min 37 s
BA
69.58
2.02
0.3225
68 min 51 s
WOA
70.34
2.93
0.3348
69 min 43 s
CSK
71.41
2.73
0.2919
69 min 06 s
Proposed IWOA
76.03
3.03
0.2395
63min32s
15
CS
69.01
2.33
0.3467
76 min 22 s
GWO
70.21
2.30
0.3029
78 min 10 s
BA
69.52
2.03
0.3220
78 min 43 s
WOA
70.28
2.94
0.3343
79 min 19 s
CSK
71.34
2.74
0.2914
78 min 56 s
Proposed IWOA
75.96
3.04
0.2390
73min27s
20
CS
69.60
2.33
0.3467
87 min 15 s
GWO
70.57
2.30
0.3029
88 min 53 s
BA
69.98
2.02
0.3220
89 min 15 s
WOA
70.67
2.93
0.3343
89 min 58 s
CSK
72.01
2.73
0.2914
89 min 39 s
Proposed IWOA
76.98
3.03
0.2390
83min46s
Best values are present in bold
Table 7
Comparison of mean accuracy, standard deviation, mean error, and execution time over ARC dataset for the different numbers of hidden nodes
Hidden node(s)
Algorithm
Mean accuracy
Standard deviation
Mean error
Execution time
5
CS
68.54
1.16
0.2084
7 min 29 s
GWO
67.99
1.07
0.2969
8 min 24 s
BA
66.11
1.68
0.2754
9 min 12 s
WOA
64.21
1.96
0.2969
9 min 02 s
CSK
67.05
1.66
0.2653
9 min 12 s
Proposed IWOA
74.83
1.89
0.1935
7min01s
7
CS
68.56
1.16
0.2072
17 min 22 s
GWO
68.00
1.07
0.2951
18 min 12 s
BA
66.11
1.67
0.2738
19 min 16 s
WOA
64.22
1.95
0.2951
19 min 12 s
CSK
67.06
1.65
0.2637
19 min 27 s
Proposed IWOA
74.73
1.88
0.1924
17min10s
9
CS
68.55
1.15
0.2064
27 min 15 s
GWO
67.99
1.06
0.2940
28 min 19 s
BA
66.11
1.66
0.2728
29 min 22 s
WOA
64.22
1.94
0.2940
29 min 19 s
CSK
67.05
1.64
0.2627
29 min 04 s
Proposed IWOA
74.72
1.87
0.1917
27min13s
11
CS
68.70
1.15
0.2064
37 min 16 s
GWO
68.15
1.06
0.2940
38 min 20 s
BA
66.26
1.66
0.2728
39 min 31 s
WOA
64.36
1.94
0.2940
39 min 17 s
CSK
67.20
1.64
0.2627
39 min 16 s
Proposed IWOA
74.89
1.87
0.1917
37min18s
13
CS
68.59
1.16
0.2067
47 min 25 s
GWO
68.03
1.07
0.2945
48 min 19 s
BA
66.15
1.67
0.2731
49 min 32 s
WOA
64.25
1.95
0.2945
49 min 20 s
CSK
67.09
1.65
0.2631
49 min 02 s
Proposed IWOA
74.76
1.88
0.1920
47min11s
15
CS
68.64
1.15
0.2069
57 min 11 s
GWO
68.09
1.06
0.2948
58 min 06 s
BA
66.20
1.66
0.2735
59 min 21 s
WOA
64.31
1.95
0.2948
59 min 11 s
CSK
67.15
1.64
0.2634
59 min 21 s
Proposed IWOA
74.83
1.88
0.1922
57min13s
20
CS
68.60
1.16
0.2070
67 min 15 s
GWO
68.04
1.06
0.2949
68 min 26 s
BA
66.16
1.67
0.2735
69 min 10 s
WOA
64.26
1.95
0.2949
69 min 28 s
CSK
67.10
1.65
0.2635
69 min 27 s
Proposed IWOA
74.78
1.88
0.1923
67min23s
Best values are present in bold
Table 8
Comparison of mean accuracy, standard deviation, mean error, and execution time over claim polarity dataset for the different numbers of hidden nodes
Hidden node(s)
Algorithm
Mean accuracy
Standard deviation
Mean error
Execution time
5
CS
71.43
0.92
0.2623
5 min 22 s
GWO
72.85
1.04
0.2477
5 min 10 s
BA
71.81
0.93
0.2584
5 min 43 s
WOA
72.85
1.33
0.2477
5 min 19 s
CSK
73.88
1.24
0.2370
4 min 56 s
Proposed IWOA
78.45
1.19
0.1898
4min27s
7
CS
71.00
0.98
0.2668
11 min 45 s
GWO
71.99
1.11
0.2566
11 min 37 s
BA
71.39
0.99
0.2627
10 min 51 s
WOA
71.99
1.42
0.2566
11 min 43 s
CSK
73.46
1.32
0.2414
10 min 26 s
Proposed IWOA
78.53
1.27
0.1891
10min12s
9
CS
71.48
0.88
0.2618
22 min 15 s
GWO
72.98
1.00
0.2463
23 min 53 s
BA
72.15
0.89
0.2549
22min15s
WOA
73.12
1.27
0.2449
22 min 58 s
CSK
73.86
1.19
0.2373
23 min 39 s
Proposed IWOA
78.31
1.14
0.1913
22 min 26 s
11
CS
71.33
0.96
0.2634
30 min 45 s
GWO
72.58
1.09
0.2505
30 min 37 s
BA
71.86
0.97
0.2580
30 min 51 s
WOA
72.64
1.39
0.2498
30 min 43 s
CSK
73.75
1.30
0.2384
30 min 06 s
Proposed IWOA
78.72
1.25
0.1870
29min32s
13
CS
71.25
1.00
0.2642
40 min 45 s
GWO
72.49
1.14
0.2514
40 min 37 s
BA
71.77
0.99
0.2588
40 min 51 s
WOA
72.56
1.45
0.2507
40 min 43 s
CSK
73.65
1.35
0.2394
40 min 06 s
Proposed IWOA
78.42
1.30
0.1901
39min32s
15
CS
71.27
0.99
0.2640
55 min 34 s
GWO
72.51
1.13
0.2512
54 min 24 s
BA
71.79
1.00
0.2586
54 min 47 s
WOA
72.58
1.44
0.2505
54 min 31 s
CSK
73.67
1.35
0.2392
55 min 31 s
Proposed IWOA
78.33
1.29
0.1911
54min30s
20
CS
71.88
0.95
0.2577
62 min 22 s
GWO
72.88
1.08
0.2474
62 min 10 s
BA
72.27
0.96
0.2537
61 min 43 s
WOA
72.98
1.38
0.2463
62 min 19 s
CSK
74.36
1.29
0.2320
61 min 56 s
Proposed IWOA
78.46
1.24
0.1897
61min27s
Best values are present in bold
Table 9
Comparison of mean accuracy, standard deviation, mean error, and execution time over perspectrum dataset for the different numbers of hidden nodes
Hidden node(s)
Algorithm
Mean accuracy
Standard deviation
Mean error
Execution time
5
CS
72.93
1.15
0.2707
7 min 15 s
GWO
72.34
1.06
0.2766
8 min 26 s
BA
70.34
1.66
0.2966
8 min 10 s
WOA
68.32
1.94
0.3168
8 min 28 s
CSK
71.34
1.64
0.2866
8 min 27 s
Proposed IWOA
79.62
1.87
0.2038
7 min 23 s
7
CS
72.95
1.15
0.2705
17 min 11 s
GWO
72.35
1.06
0.2765
18 min 06 s
BA
70.34
1.65
0.2966
18 min 21 s
WOA
68.33
1.93
0.3167
19 min 11 s
CSK
71.35
1.63
0.2865
18 min 21 s
Proposed IWOA
79.51
1.86
0.2049
17 min 13 s
9
CS
72.94
1.14
0.2706
27 min 25 s
GWO
72.34
1.05
0.2766
28 min 19 s
BA
70.34
1.64
0.2966
28 min 32 s
WOA
68.33
1.92
0.3167
29 min 20 s
CSK
71.34
1.62
0.2866
28 min 02 s
Proposed IWOA
79.50
1.85
0.2050
27 min 11 s
11
CS
73.10
1.14
0.2690
37 min 22 s
GWO
72.51
1.05
0.2749
38 min 12 s
BA
70.50
1.64
0.2950
39 min 16 s
WOA
68.48
1.92
0.3152
38 min 12 s
CSK
71.50
1.62
0.2850
39 min 27 s
Proposed IWOA
79.68
1.85
0.2032
38 min 10 s
13
CS
72.98
1.15
0.2702
47 min 15 s
GWO
72.38
1.06
0.2762
48 min 19 s
BA
70.38
1.65
0.2962
49 min 22 s
WOA
68.36
1.93
0.3164
49 min 19 s
CSK
71.38
1.63
0.2862
49 min 04 s
Proposed IWOA
79.54
1.86
0.2046
47 min 13 s
15
CS
73.03
1.14
0.2697
57 min 29 s
GWO
72.45
1.05
0.2755
58 min 24 s
BA
70.44
1.64
0.2956
58 min 12 s
WOA
68.43
1.93
0.3157
59 min 02 s
CSK
71.45
1.62
0.2855
58 min 12 s
Proposed IWOA
79.62
1.86
0.2038
57 min 01 s
20
CS
72.93
1.15
0.2707
67 min 16 s
GWO
72.34
1.05
0.2766
68 min 20 s
BA
70.34
1.65
0.2966
68 min 31 s
WOA
68.32
1.93
0.3168
69 min 17 s
CSK
71.34
1.63
0.2866
68 min 16 s
Proposed IWOA
79.62
1.86
0.2038
67 min 18 s
Best values are present in bold
Table 10
Comparison of mean accuracy, standard deviation, mean error, and execution time over snopes annotated corpus for the different numbers of hidden nodes
Hidden node(s)
Algorithm
Mean accuracy
Standard deviation
Mean error
Execution time
5
CS
53.79
1.32
0.4621
5 min 15 s
GWO
54.86
1.49
0.4514
5 min 53 s
BA
54.07
1.33
0.4593
5 min 15 s
WOA
54.86
1.90
0.4514
5 min 58 s
CSK
55.63
1.77
0.4437
4 min 39 s
Proposed IWOA
59.07
1.70
0.4093
4 min 26 s
7
CS
53.46
1.40
0.4654
11 min 45 s
GWO
54.21
1.59
0.4579
11 min 37 s
BA
53.76
1.42
0.4624
10 min 51 s
WOA
54.21
2.03
0.4579
11 min 43 s
CSK
55.32
1.89
0.4468
10 min 06 s
Proposed IWOA
59.13
1.82
0.4087
10 min 32 s
9
CS
53.82
1.26
0.4618
22 min 22 s
GWO
54.95
1.43
0.4505
23 min 10 s
BA
54.33
1.27
0.4567
22 min 43 s
WOA
55.06
1.82
0.4494
22 min 19 s
CSK
55.62
1.70
0.4438
23 min 56 s
Proposed IWOA
58.97
1.63
0.4103
22 min 27 s
11
CS
53.65
1.43
0.4635
33 min 34 s
GWO
54.58
1.63
0.4542
33 min 24 s
BA
54.04
1.42
0.4596
32 min 47 s
WOA
54.64
2.07
0.4536
32 min 31 s
CSK
55.46
1.93
0.4454
31 min 31 s
Proposed IWOA
59.05
1.86
0.4095
30 min 30 s
13
CS
53.71
1.37
0.4629
40 min 45 s
GWO
54.65
1.56
0.4535
40 min 37 s
BA
54.11
1.39
0.4589
40 min 51 s
WOA
54.70
1.99
0.4530
40 min 43 s
CSK
55.53
1.86
0.4447
40 min 26 s
Proposed IWOA
59.28
1.79
0.4072
39 min 12 s
15
CS
53.67
1.42
0.4633
55 min 22 s
GWO
54.60
1.62
0.4540
54 min 10 s
BA
54.06
1.43
0.4594
54 min 43 s
WOA
54.65
2.06
0.4535
54 min 19 s
CSK
55.47
1.93
0.4453
55 min 56 s
Proposed IWOA
58.08
1.84
0.4102
54 min 27 s
20
CS
54.13
1.36
0.4587
62 min 45 s
GWO
54.88
1.54
0.4512
62 min 37 s
BA
54.42
1.37
0.4558
61 min 51 s
WOA
54.95
1.97
0.4505
62 min 43 s
CSK
55.99
1.84
0.4401
61 min 06 s
Proposed IWOA
59.48
1.77
0.4092
61 min 32 s
Best values are present in bold
Additionally, to validate the efficacy of the proposed IWOASD, it is also compared with baseline along with state-of-the-art and recent variants of WOA such as TF-IDF, Doc2Vec + GRU-GRU, Glove + EWOA, Glove + RDWOA, etc., in terms of mean classification accuracy, mean error, standard deviation, and mean computational time. Table 11 depicts the results of all the considered methods. It is observed from the table that the proposed IWOA with Glove word embedding attains 76.53, 74.72, 78.45, 79.63, and 59.02% accuracy over FNC-1, ARC, claim polarity, perspectrum, and snopes corpus, respectively. Besides, the proposed method also outperforms other methods in terms of mean error. However, Doc2Vec + GRU-GRU shows the least variation over all the considered datasets except the FNC-1 dataset on which Tf-idf with GRU-GRU performs the best.
Table 11
Comparison of proposed IWOA-based neural network model with the state-of-the-art models
Dataset
Model
Mean accuracy (%)
Standard deviation
Mean error
Execution time
FNC-1 dataset
Tf-idf \(+\) GRU-GRU
67.17
1.34
0.3284
15 min 58 s
Doc2Vec \(+\) GRU-GRU
68.54
2.12
0.3146
18 min 09 s
Glove \(+\) GRU-GRU
68.54
2.12
0.3146
18 min 10 s
Tf-idf \(+\) SMLP
68.54
2.12
0.3146
18 min 11 s
Doc2Vec \(+\) SMLP
68.54
2.12
0.3146
18 min 09 s
AtheneMLP
71.29
1.87
0.2871
19 min 18 s
BERT base
75.84
1.49
0.2416
17 min 23 s
Glove \(+\) EWOA
73.26
3.36
0.2832
17 min 26 s
Glove \(+\) RDWOA
72.49
3.42
0.2726
16 min 24 s
Glove \(+\) NB-WOA
72.68
3.13
0.2792
17 min 29 s
Glove \(+\) proposed IWOA
76.53
3.05
0.2347
14 min 56 s
ARC dataset
Tf-idf \(+\) GRU-GRU
66.34
1.15
0.3359
8 min 35 s
Doc2Vec \(+\) GRU-GRU
65.81
1.06
0.3334
9 min 24 s
Glove \(+\) GRU-GRU
63.98
1.66
0.3721
8 min 49 s
Tf-idf \(+\) SMLP
62.15
1.94
0.3634
9 min 03 s
Doc2Vec \(+\) SMLP
64.89
1.64
0.3621
8 min 01 s
AtheneMLP
70.43
1.69
0.2967
8 min 44 s
BERT base
71.05
1.49
0.3016
8 min 19 s
Glove \(+\) EWOA
72.19
1.96
0.2943
9 min 01 s
Glove \(+\) RDWOA
71.86
2.01
0.3015
8 min 26 s
Glove \(+\) NB-WOA
71.91
2.03
0.2982
8 min 38 s
Glove \(+\) proposed IWOA
74.72
1.87
0.1913
7 min 47 s
Claim polarity dataset
Tf-idf \(+\) GRU-GRU
69.66
1.45
0.3142
5 min 12 s
Doc2Vec \(+\) GRU-GRU
70.1
0.94
0.2915
5 min 07 s
Glove \(+\) GRU-GRU
69.18
1.16
0.3126
4 min 59 s
Tf-idf \(+\) SMLP
68.26
1.28
0.3178
5 min 06 s
Doc2Vec \(+\) SMLP
71.13
1.13
0.2916
4 min 14 s
AtheneMLP
71.33
1.02
0.2866
4 min 37ec
BERT base
79.18
0.92
0.2108
5 min 11 s
Glove \(+\) EWOA
78.83
1.13
0.2432
5 min 06 s
Glove \(+\) RDWOA
76.28
0.96
0.2506
5 min 16 s
Glove \(+\) NB-WOA
77.06
1.15
0.2496
4 min 38 s
Glove \(+\) proposed IWOA
78.45
1.01
0.2198
4 min 42 s
Perspectrum dataset
Tf-idf \(+\) GRU-GRU
69.82
1.15
0.2059
8 min 35 s
Doc2Vec \(+\) GRU-GRU
70.26
1.06
0.2934
9 min 24 s
Glove \(+\) GRU-GRU
65.34
1.66
0.2721
8 min 49 s
Tf-idf \(+\) SMLP
66.41
1.94
0.2934
9 min 03 s
Doc2Vec \(+\) SMLP
70.29
1.64
0.2621
8 min 01 s
AtheneMLP
70.06
1.43
0.2991
8 min 24 s
BERT base
79.32
1.41
0.2415
8 min 38 s
Glove \(+\) EWOA
76.56
1.76
0.2365
8 min 17 s
Glove \(+\) RDWOA
79.65
1.79
0.2265
8 min 23 s
Glove \(+\) NB-WOA
78.11
1.66
0.2162
8 min 11 s
Glove \(+\) proposed IWOA
79.63
1.87
0.1913
7 min 47 s
Snopes annotated corpus
Tf-idf \(+\) GRU-GRU
51.42
1.08
0.4959
5 min 49 s
Doc2Vec \(+\) GRU-GRU
51.78
1.01
0.4912
5 min 35 s
Glove \(+\) GRU-GRU
52.76
1.53
0.4834
5 min 31 s
Tf-idf \(+\) SMLP
50.97
1.14
0.4904
6 min 09 s
Doc2Vec \(+\) SMLP
53.68
1.93
0.4648
5 min 16 s
AtheneMLP
55.49
1.46
0.4419
5 min 24 s
BERT base
59.89
1.16
0.4063
5 min 41 s
Glove \(+\) EWOA
56.32
1.36
0.4325
6 min 03 s
Glove \(+\) RDWOA
55.18
1.48
0.4215
5 min 14 s
Glove \(+\) NB-WOA
58.76
1.47
0.4401
5 min 16 s
Glove \(+\) proposed IWOA
59.02
1.72
0.4101
4 min 39 s
Best values are present in bold
Table 12
Stance detection results
Dataset
Agree
Disagree
Discuss
Unrelated
Accuracy
FNC-1 dataset
Agree
14
0
833
72
Accuracy = 76.53
Disagree
5
5
196
4
Discuss
23
2
1858
342
Unrelated
3
9
1444
7680
ARC dataset
Agree
296
67
5
28
Accuracy = 74.71
Disagree
38
332
45
30
Discuss
19
28
202
22
Unrelated
152
434
257
2493
Claim polarity dataset
Agree
289
1035
–
–
Accuracy = 78.46
Disagree
258
812
–
–
Perspectrum dataset
Agree
1195
276
–
–
Accuracy = 79.63
Disagree
289
1013
–
–
Snopes annotated corpus
Agree
3475
2703
–
–
Accuracy = 59.02
Disagree
695
1418
–
–
Best values are present in bold
Furthermore, the confusion matrix is also presented in Table 12 to know the number of correctly predicted stances by the proposed method. From the tables, it can be perceived that the proposed IWOASD shows poor performance if all four stances are considered. For the FNC-1 dataset, only 14 instances of the agreed and 5 instances of disagree stances are correctly predicted. On the contrary, the proposed IWOASD has a better recognition rate on the ARC dataset than the FNC-1 dataset. For the two-class stance datasets such as claim polarity, perspectrum, and snopes corpus, the proposed method shows much better results. The performance of the proposed IWOASD method degrades for FNC-1 and ARC dataset, since these datasets contain instances of the discussed category. The above analysis indicates that the efficiency of stance detection algorithms depends upon the number and type of stance categories in the datasets.
Conclusion and future work
This article proposes a new variant of WOA named IWOA for automated stance detection of fake news. The proposed IWOA uses tournament selection and roulette wheel selection in alternate iterations to manage the trade-off between exploration and exploitation. The proposed IWOA has been validated over 17 benchmark functions. Furthermore, an optimized neural network model based on the strength of improved whale optimization and multilayer perceptron neural network has been presented for the stance detection of fake news. The proposed optimized neural network updates weights and bias of the FFNN using an improved whale optimization algorithm. The performance of the proposed model has been tested on five stance detection datasets and compared with CS, GWO, BA, WOA, and CSK. The proposed IWOA-optimized FFNN model achieves the highest accuracy as compared to the other considered models. Moreover, the proposed model also outperforms other considered models for the performance measures mean error and execution time for more than 80% of the datasets. The proposed IWOA-optimized neural network shows better results; however, improvement is still required. The proposed IWOA employs random choices to find the optimal solution, which means that the computing time and the solution quality are actually random variables. Due to this stochastic nature, its rigorous analysis would be very difficult. Moreover, it cannot be guaranteed that the solution found in different runs will be globally optimal or of high quality. To mitigate the same, IWOA has been executed 30 times and its mean value has been used for comparison. From the results, it has been found that the proposed IWOA shows much better performance than other algorithms on single-objective problems. However, for the multi-objective problems, the proposed IWOA and other algorithms show the same performance. In future work, feature selection methods and different optimization methods can be explored for improving the accuracy. Furthermore, deep learning models such as CNN, LSTM, and BiLSTM could be also investigated. Besides, transfer learning and multitask methodologies could be also considered to exploit knowledge from other related domains.
Acknowledgements
The author is also thankful to Dr. Himanshu Mittal, Assistant Professor at Jaypee Institute of Information Technology, Noida, India for tendering his valuable help for implementing the proposed method.
Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
