Linguistic modifiers such as negators (e.g., not), intensifiers (e.g., very) and modals (e.g., would) are commonly used in expressing opinions. These modifiers play an important role in recognizing the sentiment intensity of multi-word phrases because they may lead to an intensity shift and polarity reversal for the words they modify. Appropriately modeling the effect of such modifiers on the intensity shift can greatly improve the performance of phrase-level sentiment intensity prediction. To this end, this paper proposes two neural network (NN) models organized in a pipelined fashion to determine 1) the intensity of individual words and 2) the shift weights of modifiers representing the degrees of intensity change for the words they modify. The intensity of a phrase can then be determined by combining the intensity of the constituent word and the shift weight of the modifier within the phrase. When measuring the word intensity, the first NN model introduces a hidden layer as a filter to select appropriate similar seed words in the prediction process. Automatic word intensity prediction can address the unknown intensities of words not covered in sentiment lexicons. In learning the modifier weights, the second NN model considers both the weights of individual modifiers and groups of modifiers to capture various intensity shift effects caused by them. Experiments on a SemEval-2016 dataset showed that the proposed method yielded better prediction performance for both single words and multi-word phrases.

Dimensional sentiment analysis aims to recognize continuous numerical values in multiple dimensions such as the valence-arousal (VA) space. Compared to the categorical approach that focuses on sentiment classification such as binary classification (i.e., positive and negative), the dimensional approach can provide a more fine-grained sentiment analysis. This article proposes a tree-structured regional CNN-LSTM model consisting of two parts: regional CNN and LSTM to predict the VA ratings of texts. Unlike a conventional CNN which considers a whole text as input, the proposed regional CNN uses a part of the text as a region, dividing an input text into several regions such that the useful affective information in each region can be extracted and weighted according to their contribution to the VA prediction. Such regional information is sequentially integrated across regions using LSTM for VA prediction. By combining the regional CNN and LSTM, both local (regional) information within sentences and long-distance dependencies across sentences can be considered in the prediction process. To further improve performance, a region division strategy is proposed to discover task-relevant phrases and clauses to incorporate structured information into VA prediction. Experimental results on different corpora show that the proposed method outperforms lexicon-, regression-, conventional NN and other structured NN methods proposed in previous studies.

Danger-level analysis is widely used to prevent potential driving risks based on driving performance. Such analysis is essential for monitoring driver performance. Moreover, danger-level analysis is vital for automotive safety systems and driving assistance applications. However, danger-level analysis that simultaneously considers driver-, vehicle-, and road-related information from driving data has rarely been conducted. Such analysis is very challenging due to the issues associated with the high volume and high variety in multisourced driving data. In this paper, we propose a novel danger-level analysis framework for dealing with high variety and high volume problems of multisourced driving data. Built upon a feature extraction method in the proposed framework, we first profile multisourced driving features for overcoming the variety problem. Next, danger-level analysis is formulated as a multiobjective pursuit problem in a linear model. The problem is then solved using a semisupervised learning strategy to overcome the volume issue. Therefore, the danger level can be accurately estimated from multisourced driving data by using the proposed framework. The experimental results indicate that the proposed framework outperforms existing machine learning techniques for multisourced driving data.

Jin Wang, Liang-Chih Yu, K. Robert Lai, Xuejie Zhang. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Word embeddings that provide continuous low-dimensional vector representations of words have been extensively used for various natural language processing tasks. However, existing context-based word embeddings such as Word2vec and GloVe typically fail to capture sufficient sentiment information, which may result in words with similar vector representations having an opposite sentiment polarity (e.g., good and bad), thus degrading sentiment analysis performance. To tackle this problem, recent studies have suggested learning sentiment embeddings to incorporate the sentiment polarity (positive and negative) information from labeled corpora. This study adopts another strategy to learn sentiment embeddings. Instead of creating a new word embedding from labeled corpora, we propose a word vector refinement model to refine existing pretrained word vectors using real-valued sentiment intensity scores provided by sentiment lexicons. The idea of the refinement model is to improve each word vector such that it can be closer in the lexicon to both semantically and sentimentally similar words (i.e., those with similar intensity scores) and further away from sentimentally dissimilar words (i.e., those with dissimilar intensity scores). An obvious advantage of the proposed method is that it can be applied to any pretrained word embeddings. In addition, the intensity scores can provide more fine-grained (real-valued) sentiment information than binary polarity labels to guide the refinement process. Experimental results show that the proposed refinement model can improve both conventional word embeddings and previously proposed sentiment embeddings for binary, ternary, and fine-grained sentiment classification on the SemEval and Stanford Sentiment Treebank datasets.

Word embeddings that can capture semantic and syntactic information from contexts have been extensively used for various natural language processing tasks.However, existing methods for learning contextbased word embeddings typically fail to capture sufficient sentiment information.This may result in words with similar vector representations having an opposite sentiment polarity (e.g., good and bad), thus degrading sentiment analysis performance.Therefore, this study proposes a word vector refinement model that can be applied to any pre-trained word vectors (e.g., Word2vec and GloVe).The refinement model is based on adjusting the vector representations of words such that they can be closer to both semantically and sentimentally similar words and further away from sentimentally dissimilar words.Experimental results show that the proposed method can improve conventional word embeddings and outperform previously proposed sentiment embeddings for both binary and fine-grained classification on Stanford Sentiment Treebank (SST).

Protein ubiquitination, involving the conjugation of ubiquitin on lysine residue, serves as an important modulator of many cellular functions in eukaryotes. Recent advancements in proteomic technology have stimulated increasing interest in identifying ubiquitination sites. However, most computational tools for predicting ubiquitination sites are focused on small-scale data. With an increasing number of experimentally verified ubiquitination sites, we were motivated to design a predictive model for identifying lysine ubiquitination sites for large-scale proteome dataset. This work assessed not only single features, such as amino acid composition (AAC), amino acid pair composition (AAPC) and evolutionary information, but also the effectiveness of incorporating two or more features into a hybrid approach to model construction. The support vector machine (SVM) was applied to generate the prediction models for ubiquitination site identification. Evaluation by five-fold cross-validation showed that the SVM models learned from the combination of hybrid features delivered a better prediction performance. Additionally, a motif discovery tool, MDDLogo, was adopted to characterize the potential substrate motifs of ubiquitination sites. The SVM models integrating the MDDLogo-identified substrate motifs could yield an average accuracy of 68.70 percent. Furthermore, the independent testing result showed that the MDDLogo-clustered SVM models could provide a promising accuracy (78.50 percent) and perform better than other prediction tools. Two cases have demonstrated the effective prediction of ubiquitination sites with corresponding substrate motifs.

Compared to the categorical approach that represents affective states as several discrete classes (e.g., positive and negative), the dimensional approach represents affective states as continuous numerical values in multiple dimensions, such as the valence-arousal (VA) space, thus allowing for more fine-grained sentiment analysis. In building dimensional sentiment applications, affective lexicons with VA ratings are useful resources but are still very rare. Several semi-supervised methods such as the kernel method, linear regression, and the pagerank algorithm have been investigated to automatically determine the VA ratings of affective words from a set of semantically similar seed words. These methods suffer from two major limitations. First, they apply an equal weight to all seeds similar to an unseen word in predicting its VA ratings. Second, even similar seeds may have quite different ratings (or an inverse polarity) of valence/arousal to the unseen word, thus reducing prediction performance. To overcome these limitations, this study proposes a community-based weighted graph model that can select seeds which are both similar to and have similar ratings (or the same polarity) with each unseen word to form a community (subgraph) so that its VA ratings can be estimated from such high-quality seeds using a weighted propagation scheme. That is, seeds more similar to unseen words contribute more to the estimation process. Experimental results show that the proposed method yields better prediction performance for both English and Chinese datasets.

Text-based sentiment analysis is a growing research field in affective computing, driven by both commercial applications and academic interest. Continuous dimensional representations, such as valence-arousal (VA) space, can represent the affective state more precisely than discrete effective representations. In building dimensional sentiment applications, affective lexicons with valence-arousal ratings are useful resources but are still very rare. Therefore, recent studies have investigated the automatic development of VA lexicons using linear regression techniques. One of the major limitations of linear regression is the under-fitting problem which can cause a poor fit between the algorithm and the training data. To tackle this problem, this study proposes the use of a locally weighted linear regression (LWLR) model to predict the valence-arousal ratings of affective words. The locally weighted method performs a regression around the point of interest using only training data that are "local" to that point, and thus can reduce the impact of noise from unrelated training data. Experimental results show that the proposed method achieved better performance for VA word prediction.

MicroRNAs (miRNAs) are a group of small noncoding RNA (ncRNA) molecules that play an important role in biological functions. This paper proposes a microRNA prediction application, which is based on the multi-layer hierarchical MapReduce framework and provides four prediction workflows for four different datasets: miRNA-like sequences, miRNA cluster sequences, unknown miRNA sequences and the next generation sequencing (NGS) sequences. These workflows include four core procedures for finding the genome location, applying the biological filtering criteria and a genetic algorithm based pre-miRNA classifier. Each procedure works as a MapReduce task and uses JSON format to translate the MapReduce output to the next MapReduce procedure. Experimental results show that the proposed miRNA predictor not only achieves high sensitivity and accuracy, but also has ability to process more than million sequences in acceptable time by relying on the multi-layer hierarchical MapReduce framework.

Since typical introductory programming curricula emphasize the features and syntax of a programming language, rather than the application of programming strategies, most novices are incapable of solving programming problems at the end of curricula. The difference between experts and novices is that experts can develop abstract representation of solutions, called the programming plans, for problems through the experiences obtained from past problems and solutions. The experts then use the programming plans to solve similar program problems. In this study, we develop a system to help students comprehend programming plans by explaining program codes as well as investigate the relationship between goal orientation patterns and self-efficacy.

BACKGROUND: In eukaryotes, ubiquitin-conjugation is an important mechanism underlying proteasome-mediated degradation of proteins, and as such, plays an essential role in the regulation of many cellular processes. In the ubiquitin-proteasome pathway, E3 ligases play important roles by recognizing a specific protein substrate and catalyzing the attachment of ubiquitin to a lysine (K) residue. As more and more experimental data on ubiquitin conjugation sites become available, it becomes possible to develop prediction models that can be scaled to big data. However, no development that focuses on the investigation of ubiquitinated substrate specificities has existed. Herein, we present an approach that exploits an iteratively statistical method to identify ubiquitin conjugation sites with substrate site specificities. RESULTS: In this investigation, totally 6259 experimentally validated ubiquitinated proteins were obtained from dbPTM. After having filtered out homologous fragments with 40% sequence identity, the training data set contained 2658 ubiquitination sites (positive data) and 5532 non-ubiquitinated sites (negative data). Due to the difficulty in characterizing the substrate site specificities of E3 ligases by conventional sequence logo analysis, a recursively statistical method has been applied to obtain significant conserved motifs. The profile hidden Markov model (profile HMM) was adopted to construct the predictive models learned from the identified substrate motifs. A five-fold cross validation was then used to evaluate the predictive model, achieving sensitivity, specificity, and accuracy of 73.07%, 65.46%, and 67.93%, respectively. Additionally, an independent testing set, completely blind to the training data of the predictive model, was used to demonstrate that the proposed method could provide a promising accuracy (76.13%) and outperform other ubiquitination site prediction tool. CONCLUSION: A case study demonstrated the effectiveness of the characterized substrate motifs for identifying ubiquitination sites. The proposed method presents a practical means of preliminary analysis and greatly diminishes the total number of potential targets required for further experimental confirmation. This method may help unravel their mechanisms and roles in E3 recognition and ubiquitin-mediated protein degradation.

In this paper, we describe a work-in-progress research to develop a program control structure visualization tool for transforming student programs into graphical representations of control structure, including sequences, selections, and iterations. The tool can be used for assisting novice students in debugging their programs by checking whether the transformed control structure of their programs is consistent with their program plans. In addition, the graphical representation of student programs can be compared to that of a model program to detect errors in program plans.

This paper proposes a visualization tool for supporting representation construction via constraint-based approach. A physics problem can be represented as an operational concept map (OCM) by identifying important concepts and relationships among these concepts from the problem scenario. This form of representation enables the visualization tool to assist a solver to construct a graphical, simulation-enabled concept map regarding a specific physics problem solving. To help solvers construct problem representation and validate its corresponding hypothetical relationships, an OCM integrates the organization function of concept map with hypothesis simulation capability. The application of an OCM is further introduced with a kinematics problem in order to demonstrate the features of hypothesis creations, evaluation, and revision with feedback messages.

This study proposes a negotiation-based approach to combine the notion of adaptivity and adaptability for adaptive learning system. The system suggests adaptation and the student also submits his/her preference of adaptation. While the student preference violates the system suggestion, the student needs to negotiate with the system to reach an agreement of adaptation. To delineate the proposed framework, an undergraduate Java programming course is provided to illustrate how the negotiation-based adaptive learning system can support the generation of personalized adaptive learning sequences.

In this study, we explored the students’ computer-supported collaborative learning behavior based on the Facebook platform. Sixty two senior college students major in Information Management took Decision Support System (DSS) class. Besides the lectures and class discussion, the students participated in the DSS Facebook for collaborative learning. We found that students’ characteristics (e.g., gender and mindset of learning) are important factors to affect their Facebook usage behavior and learning performance. The students using DSS Facebook more often get better performance in their final projects, learning satisfaction and the online communication behavior survey. We also found that gender affects the usage of social networks platform. For instance, male students use social networks platform several times per week and get better performance in online communication, learning satisfaction and creativity self-efficacy.

This study presents a conceptual framework for providing intelligent supports through agent negotiation, fuzzy constraints and natural language processing to enhance the effectiveness of peer assessment. By using fuzzy constraints, it not only provides a flexible marking scheme to deal with the imprecision and uncertainty for the representation of assessment. Additionally, a fuzzy constraint-based negotiation mechanism is employed to coordinate the cognitive differences between students. Through iterative agent negotiation, students can reconcile the differences and reach an agreement on the assessment results. Owing to the difficulty in reading fuzzy sets of assessment results we incorporate the technique of linguistic approximation to translate fuzzy sets into natural language to facilitate students to understand assessment feedback. Experimental results indicated that students were able to acquire more meaningful and readable feedback to reflect upon and revise their work.

Problem-based learning (PBL) encourages learners to acquire new knowledge through cognitive model development and hypotheses-deduction process. Graphical modeling tools, such as concept map, can help learners to exteriorize the process of knowledge construction and visualize learner's cognitive model to support meaningful learning. But few of those tools integrate hypothesis-reasoning mechanism into the process to support the discovery of knowledge deficiencies and foster in-depth reflection. To that end, this paper proposes a novel graphical tool, operational concept map (OCM), for supporting the PBL. In OCM, it combines the merits of conceptual modeling tool and computer-based simulation. The OCM not only inherits all basic functions of concept map, also facilitates the learning-by-doing approach for problem solving. Application of the proposed idea is illustrated in learning a well-known sort algorithm, bubble sort, in the sort problem.

This paper presents a multi-hop dynamic moving algorithm to repair the coverage holes, while maintaining the initial topology and connectivity, for the wireless sensor networks. In such environment, the movement of each sensor node is fully distributed and autonomous. The proposed algorithm takes the density measures into consideration for determining which node to move and to what direction and distance iteratively to heal the holes and also to maximize the overall coverage area. Simulation results suggest that the proposed algorithm outperforms other maintenance algorithms in terms of coverage, convergence speed, and energy consumption.

In a wireless ad hoc network, multiple end-to-end flows that traverse through the same wireless communication range compete for the shared wireless bandwidth. Then, how to allocate those shared bandwidth among flows is one of most critical issues for the system performance. To that end, this paper proposes an auction-based bandwidth allocation with multi-hop flow coordination mechanism to enhance various performance measures. Experimental results suggest that the proposed approach outperforms other approaches in terms of network throughput, bandwidth utilization, fairness, end-to-end delay, and packet loss rate.

This paper presents a conceptual framework of learning through agent negotiation. In a computer-supported learning environment, interaction plays a critical role for developing instructional strategies and promoting learning effectiveness. Learners, peers, and instructors often bring different ideas and perspectives to the learning process. Thus, how to resolve these differences and without creating unnecessary discontentment can be of paramount important in a learning system. This study aims at developing a unified conceptual framework to promote effective interaction through negotiation and then to enhance the learning performance as well.

This paper integrates the concept of agent negotiation into adaptive learning to support effective learner-instructor interaction. It intends to capture the interactions that are real between learner and instructor, but often simplified or even ignored in most adaptive learning approaches. An iterative negotiation protocol is employed to reconcile mutual thinking and cognitive differences between learner and instructor and then to dynamically produce the most appropriate learning sequences. To delineate the proposed framework, an example relevant to Java programming course is provided to illustrate how the adaptive learning environment with negotiation mechanism can improve the effectiveness of learning.

This paper presents a business gaming simulator via fuzzy constraints-directed approach for supporting collaborative problem-based learning (PBL). In this gaming simulator, the problem is formulated and analyzed as a distributed fuzzy constraint satisfaction problem (DFCSP) to assist students realizing a wide range of business scenarios from different perspectives. The structure of the problem is represented as decision networks to help students to explore rdquowhat-ifrdquo hypotheses as well as to allow instructor to give appropriate instructions. In addition, a fuzzy constraint-directed agent negotiation mechanism is also provided to mimic more closely and naturally with the spirit of the real-world problem-solving for collaborative strategic planning and decision making. To demonstrate the ideas of the proposed framework for supporting PBL, we have developed a prototyped business simulator, MANAGER, and applied it successfully in the beer game for supporting collaborative PBL.

This paper presents a novel motion detection scheme by using the momentum of received signal sstrength (MRSS) to improve the quality of handoff in a general wireless network. MRSS can detect the motion state of a mobile node (MN) without assistance of any positioning service. Although MRSS is sensitive in detecting user's motion, it is static and fails to detect quickly the motion changes of users. Thus, a novel motion state dependent MRSS scheme called dynamic MRSS (DMRSS) algorithm is proposed to address this issue. Extensive simulation experiments were conducted to study performance of our presented algorithms. The simulation results show that MRSS and DMRSS can be used to assist a handoff algorithm in substantially reducing unnecessary handoff and saving power.

This paper presents a fuzzy constraint-directed negotiation mechanism for agent-based scheduling. Scheduling problem is modeled as a set of fuzzy constraint satisfaction problems (FCSP), interlinked together by inter-agent constraints. Each FCSP represents the perspective of participants and is governed by agents. Negotiation process is considered as a global consistency enforcing via iterative constraint adjustment and relaxation. To facilitate convergence and improve solution quality for a particular performance measure, sharing meta-scheduling information during negotiation is applied. Experimental results suggest that the proposed approach not only can obtain a high quality schedule in a cost-effective manner, but also provides superior performance in all criteria to other negotiation models for agent-based scheduling.

Incorporation of various sources of a priori knowledge into data-driven fuzzy modelling is an important task. But a major problem with current approaches is that they are mostly problem-specific and lacking an effective framework to bring different sources of knowledge into the task of modelling. In this paper, we propose a constraint-based framework for the incorporation of a priori knowledge into data-driven-based fuzzy modelling. We first investigate a logical taxonomy of background knowledge in learning a fuzzy model. Then, based on this taxonomy, we can develop a framework for incorporating prior knowledge into a constraint-based fuzzy modelling. Finally, two simulation examples, a nonlinear function fitting problem and a dynamic time series prediction problem, are provided for the embodiment of the proposed idea.

This paper presents a belief learning model for fuzzy constraint-directed agent negotiation. The main features of the proposed model include: 1) fuzzy probability constraints for increasing the efficiency on the convergence of behavior patterns, and eliminating the noisy hypotheses or beliefs, 2) fuzzy instance matching method for reusing the prior opponent knowledge to speed up the problem-solving, and inferring the proximate regularities to acquire a desirable result on forecasting opponent behavior, and 3) adaptive interaction for making a dynamic concession to fulfill a desirable objective. Experimental results suggest that the proposed framework can improve both negotiation qualities.

A negotiation-based intelligent tutoring system of which the tutor and learner agents in playing problem-specific roles is developed to assist learner in problem solving through a negotiation mechanism. During the problem solving, fuzzy constraint-directed negotiation is employed to coordinate the mutual thinking between tutor and learner agents, and it can resolve the cognition differences and thereby reach a consensus more effectively. To embody the idea of the proposed methodology, a real-world example relevant to the software cost estimation is described to illustrate how a negotiation can be used for supporting the problem solving in a tutoring learning environment.

This paper presents a fuzzy constraint-directed approach to the development of agent-based business simulator for supporting collaborative learning. The advantages of this approach can be summarized in two folds: it provides (1) a general purpose framework for the knowledge representation and problem solving in business simulation, and (2) a fully distributed computational model along with a negotiation algorithm that can be easily built to mimic more closely to the real-world scenario of collaborative strategic planning and decision making. To demonstrate the usefulness of the proposed framework, we have prototyped a business simulator, MANAGE, and applied successfully to build a beer game for collaborative learning.

In peer assessment, the awarding grades might not accurately reflect the students' achievement due to potential rating bias or individual abilities. The proposed methodology aims at aggregating students' ratings to reduce personal bias using agent negotiation. We consider individual learning styles of assessors into the negotiation process and show by an illustrative example and an experiment how the accuracy of assessment results can be improved through incorporating learning styles. The more accurate feedback provides students a better quality of assessment which enables them to reflect their effort and abilities.

This study presents a negotiation-based e-learning framework to promote learning motive and enhance the interaction between learners and instructors. In terms of learning communication, a fuzzy constraint-based agent negotiation mechanism is proposed as a flexible model to suggest pedagogies or learning strategies. Examples are provided to illustrate the proposed framework for constructive and collaborative learning.

This paper presents a novel methodology for modeling collaborative learning as multi-issue agent negotiation using fuzzy constraints. Agent negotiation is an iterative process through which the proposed methodology can aggregate students' marking to reduce personal bias. In the framework, students define individual fuzzy membership functions based on their evaluation concepts and agents stand for students to negotiate with each other in the assessment process. By applying the method, agents can reach a mutually acceptable agreement to overcome the subjective judgment and the unfair assessment. Thus, the negotiated agreement provides students a better quality of assessment and enables students to improve the learning effectiveness. An example application to negotiate the assessment among three students is provided to illustrate the assessment process of the proposed framework.

In e‐business, disputes between two or more parties arise for various reasons and involve different issues. Thus, resolution of these disputes frequently relies on some form of negotiation. This article presents a general problem‐solving framework for modeling multi‐issue multilateral agent negotiation using fuzzy constraints in e‐business. Fuzzy constraints are thus used not only to define each agent's demands involving human concepts, but also to represent the relationships among agents. A concession strategy, based on fuzzy constraint‐based problem‐solving, is proposed to relax demands, and a trade‐off strategy is presented to evaluate existing alternatives. This approach provides a systematic method for reaching an agreement that benefits all agents with a high satisfaction degree of constraints. Meanwhile, by applying the method, agents can move toward an agreement more quickly, because their search focuses only on the feasible solution space. An example application to negotiate an insurance policy among agents is provided to demonstrate the usefulness and effectiveness of the proposed framework.

This paper proposes a general problem-solving framework for modeling of agent negotiation via fuzzy constraint processing to acquire an optimal solution that satisfies agents in the environment of incomplete and imprecise information. In this approach, fuzzy constraint processing serves not only to improve the efficiency of the agent's individual negotiation strategy which aims at the maximal satisfaction of its self-interest, but also to represent a negotiation protocol which determines legal or meaningful sequences of messages that must be satisfied. Thus, our approach is to model agent negotiation based on fuzzy constraint processing as a distributed fuzzy constraint satisfaction problem which objective is to find a solution that has maximal satisfaction for all fuzzy constraints in the distributed environment. In this paper, we introduce the concept of fuzzy constraint processing and the definition of distributed fuzzy constraint network. Then, the approach for modeling agent negotiation via fuzzy constraint processing is presented. Next, to demonstrate the effectiveness of this approach, an example of phone-trading in electronic marketplace is provided.

In this paper, we present the design of a constraint-based fuzzy logic controller (CFLC) for pH control of a flue gas desulfurization (FGD) system. Based on the concept of a CFLC, the control strategies for the FGD process are modeled as a network; the control actions are then governed to the intended states of satisfying all the constraints in the network. Additionally, we also include the simulation results to show the effectiveness of CFLC for the FGD system, while it has not been successfully dealt with through either the conventional PID or rule-based fuzzy logic approach.