Pub Date : 2018-10-01DOI: 10.1109/BRACIS.2018.00053
Marcelo Pita, G. Pappa
Short texts are present in many computer systems. Examples include social media messages, advertisement, Q&A websites, and an increasing number of other applications. They are characterized by little context words and a large vocabulary. As a consequence, traditional short text representations, such as TF and TF-IDF, have high dimensionality and are very sparse. The research field of word vectors has produced interesting word representations that are discriminative regarding semantics, which can be algebraically composed to create vector representations for paragraphs and documents. Literature reports limitations of this approach, producing the alternative Paragraph Vector method. Firstly, we investigate whether these limitations involving word vector operations are true for short text. Then, we propose a novel representation method based on the PSO meta-heuristic. Results in a document classification task are competitive with TF-IDF and show significant improvement over Paragraph Vector, with the advantage of dense and compact document vector representation.
{"title":"Strategies for Short Text Representation in the Word Vector Space","authors":"Marcelo Pita, G. Pappa","doi":"10.1109/BRACIS.2018.00053","DOIUrl":"https://doi.org/10.1109/BRACIS.2018.00053","url":null,"abstract":"Short texts are present in many computer systems. Examples include social media messages, advertisement, Q&A websites, and an increasing number of other applications. They are characterized by little context words and a large vocabulary. As a consequence, traditional short text representations, such as TF and TF-IDF, have high dimensionality and are very sparse. The research field of word vectors has produced interesting word representations that are discriminative regarding semantics, which can be algebraically composed to create vector representations for paragraphs and documents. Literature reports limitations of this approach, producing the alternative Paragraph Vector method. Firstly, we investigate whether these limitations involving word vector operations are true for short text. Then, we propose a novel representation method based on the PSO meta-heuristic. Results in a document classification task are competitive with TF-IDF and show significant improvement over Paragraph Vector, with the advantage of dense and compact document vector representation.","PeriodicalId":405190,"journal":{"name":"2018 7th Brazilian Conference on Intelligent Systems (BRACIS)","volume":"157 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129697995","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/BRACIS.2018.00082
D. Tozadore, C. M. Ranieri, Guilherme V. Nardari, V. Guizilini, R. Romero
Understanding people's emotions may be important to achieve success in behavior adaptability and, consequently, to sustain long-term human-robot interactions. Most emotion recognition systems consist in classifying a given input into one out of seven basic emotions, following Ekman's model. However, it is sometimes enough for the customization of a robot's behavior to recognize whether an emotion is positive or negative, in order to approach more often subjects which display more positive emotional reactions. In this article, two approaches to that effect are proposed and compared. The first one, named pre-grouping, refers to combining the four negative emotions into one single class and use it to train a classifier. The second one, named post-grouping, refers to applying classifiers to classify the seven basic emotions and interpret their negative outputs as related to a single class. Furthermore, a novel dataset entitled QIDER, based on queries in a search engine and well defined facial cues, is introduced and made available for public use. Both approaches led to more balanced precision scores among all classes, which may make them a suitable choice for applications in human-robot interaction. Several experiments have been performed and post-grouping is shown to produce better overall accuracy.
{"title":"Effects of Emotion Grouping for Recognition in Human-Robot Interactions","authors":"D. Tozadore, C. M. Ranieri, Guilherme V. Nardari, V. Guizilini, R. Romero","doi":"10.1109/BRACIS.2018.00082","DOIUrl":"https://doi.org/10.1109/BRACIS.2018.00082","url":null,"abstract":"Understanding people's emotions may be important to achieve success in behavior adaptability and, consequently, to sustain long-term human-robot interactions. Most emotion recognition systems consist in classifying a given input into one out of seven basic emotions, following Ekman's model. However, it is sometimes enough for the customization of a robot's behavior to recognize whether an emotion is positive or negative, in order to approach more often subjects which display more positive emotional reactions. In this article, two approaches to that effect are proposed and compared. The first one, named pre-grouping, refers to combining the four negative emotions into one single class and use it to train a classifier. The second one, named post-grouping, refers to applying classifiers to classify the seven basic emotions and interpret their negative outputs as related to a single class. Furthermore, a novel dataset entitled QIDER, based on queries in a search engine and well defined facial cues, is introduced and made available for public use. Both approaches led to more balanced precision scores among all classes, which may make them a suitable choice for applications in human-robot interaction. Several experiments have been performed and post-grouping is shown to produce better overall accuracy.","PeriodicalId":405190,"journal":{"name":"2018 7th Brazilian Conference on Intelligent Systems (BRACIS)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125423467","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/BRACIS.2018.00036
L. M. Pavelski, Marie-Éléonore Kessaci, M. Delgado
This work proposes a meta-learning system based on Gradient Boosting Machines to recommend local search heuristics for solving flowshop problems. The investigated approach can decide if a metaheuristic (MH) is suitable for each instance. It can also provide well-suited parameters for each recommended MH using data from Irace parameter tuning. This paper considers four MHs (Hill Climbing, Tabu Search, Simulated Annealing, and Iterated Local Search) as candidates to solve several flowshop instances. In the experiments, 540 flowshop problems (with different sizes, variants, and objectives) and 50 instances for each problem are considered, resulting in a total of 27,000 instances being addressed. We use simple low-level meta-features in the meta-learning system like the number of jobs and machines, processing time distribution, flowshop variant, objective, and evaluations budget. Besides testing the recommendations in terms of accuracy and Kappa (for MH and categorical parameters), RMSE and R2 (for numerical parameters), we also explore the importance of each meta-feature in MH recommendation models. Moreover, we perform a multiple correspondence analysis on MH configurations to gain further insights into the parameters values. Results show that the proposed approach is promising, particularly for MH recommendation.
{"title":"Recommending Meta-Heuristics and Configurations for the Flowshop Problem via Meta-Learning: Analysis and Design","authors":"L. M. Pavelski, Marie-Éléonore Kessaci, M. Delgado","doi":"10.1109/BRACIS.2018.00036","DOIUrl":"https://doi.org/10.1109/BRACIS.2018.00036","url":null,"abstract":"This work proposes a meta-learning system based on Gradient Boosting Machines to recommend local search heuristics for solving flowshop problems. The investigated approach can decide if a metaheuristic (MH) is suitable for each instance. It can also provide well-suited parameters for each recommended MH using data from Irace parameter tuning. This paper considers four MHs (Hill Climbing, Tabu Search, Simulated Annealing, and Iterated Local Search) as candidates to solve several flowshop instances. In the experiments, 540 flowshop problems (with different sizes, variants, and objectives) and 50 instances for each problem are considered, resulting in a total of 27,000 instances being addressed. We use simple low-level meta-features in the meta-learning system like the number of jobs and machines, processing time distribution, flowshop variant, objective, and evaluations budget. Besides testing the recommendations in terms of accuracy and Kappa (for MH and categorical parameters), RMSE and R2 (for numerical parameters), we also explore the importance of each meta-feature in MH recommendation models. Moreover, we perform a multiple correspondence analysis on MH configurations to gain further insights into the parameters values. Results show that the proposed approach is promising, particularly for MH recommendation.","PeriodicalId":405190,"journal":{"name":"2018 7th Brazilian Conference on Intelligent Systems (BRACIS)","volume":"70 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123527061","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/BRACIS.2018.00017
Rafaella F. Vale, R. Lins, Rafael Ferreira
Automatic text summarization is proving itself useful to sieve relevant content from the Internet and digital libraries with reduced human effort. Nevertheless, extractive summarization approaches have limitations, possibly not fully capturing the informativeness of a text. A potential strategy to address this problem is the adoption of sentence simplification methods. This work focuses on the evaluation of sentence simplification methods as a preprocessing step for extractive text summarization in order to answer the question of whether sentence simplification increases the informativeness of extractive summaries. Four different sentence simplification methods, two being simple filters and the other two performing rule-based transformations, are assessed here in order to point out the best method for such a purpose. Fifteen sentence scoring methods for summarization are applied in combination with the simplification methods to a corpus of 1,038 news articles in English. The results suggest that the transformation approaches, which take into account linguistic features and grammaticality, achieve the best performance.
{"title":"Assessing Sentence Simplification Methods Applied to Text Summarization","authors":"Rafaella F. Vale, R. Lins, Rafael Ferreira","doi":"10.1109/BRACIS.2018.00017","DOIUrl":"https://doi.org/10.1109/BRACIS.2018.00017","url":null,"abstract":"Automatic text summarization is proving itself useful to sieve relevant content from the Internet and digital libraries with reduced human effort. Nevertheless, extractive summarization approaches have limitations, possibly not fully capturing the informativeness of a text. A potential strategy to address this problem is the adoption of sentence simplification methods. This work focuses on the evaluation of sentence simplification methods as a preprocessing step for extractive text summarization in order to answer the question of whether sentence simplification increases the informativeness of extractive summaries. Four different sentence simplification methods, two being simple filters and the other two performing rule-based transformations, are assessed here in order to point out the best method for such a purpose. Fifteen sentence scoring methods for summarization are applied in combination with the simplification methods to a corpus of 1,038 news articles in English. The results suggest that the transformation approaches, which take into account linguistic features and grammaticality, achieve the best performance.","PeriodicalId":405190,"journal":{"name":"2018 7th Brazilian Conference on Intelligent Systems (BRACIS)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126483878","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/BRACIS.2018.00072
Vagner de Oliveira Gabriel, D. Adamatti, Alison R. Panisson, Rafael Heitor Bordini, C. Z. Billa
The theory of argumentation spans several fields of knowledge, gaining significant space in the community of multiagent systems because it gives support for agents to reason about uncertain beliefs. This work describes the development of an argumentation-based inference architecture for BDI agents, which was developed based on Toulmin's model of argumentation. The philosopher Stephen Toulmin claimed that arguments typically consist of six parts: data, warrants, claim, backing, qualifiers, and rebuttals. Using the proposed architecture, an agent is able to create new beliefs based on available evidence and to justify such beliefs.
{"title":"Argumentation-Based Reasoning in BDI Agents Using Toulmin's Model","authors":"Vagner de Oliveira Gabriel, D. Adamatti, Alison R. Panisson, Rafael Heitor Bordini, C. Z. Billa","doi":"10.1109/BRACIS.2018.00072","DOIUrl":"https://doi.org/10.1109/BRACIS.2018.00072","url":null,"abstract":"The theory of argumentation spans several fields of knowledge, gaining significant space in the community of multiagent systems because it gives support for agents to reason about uncertain beliefs. This work describes the development of an argumentation-based inference architecture for BDI agents, which was developed based on Toulmin's model of argumentation. The philosopher Stephen Toulmin claimed that arguments typically consist of six parts: data, warrants, claim, backing, qualifiers, and rebuttals. Using the proposed architecture, an agent is able to create new beliefs based on available evidence and to justify such beliefs.","PeriodicalId":405190,"journal":{"name":"2018 7th Brazilian Conference on Intelligent Systems (BRACIS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133510764","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/BRACIS.2018.00083
J. Andrade, Robson Oliveira, T. Silva, J. Maia, G. Campos
In the CTO problem a group of moving observers should monitor a group of moving targets in order to maximize the average number of observed targets (ANOT). This work proposes and evaluates a centralized strategy based on FCM (Fuzzy C-means) to determine the movement of the observers in a setting in which the targets move randomly. The observer group is modeled as an agent organization specified in an FSB (Function-Structure-Behavior) standard. Compared with similar works, the test results are competitive in addition to avoiding known drawbacks reported in the literature.
{"title":"Organization/fuzzy Approach to the CTO Problem","authors":"J. Andrade, Robson Oliveira, T. Silva, J. Maia, G. Campos","doi":"10.1109/BRACIS.2018.00083","DOIUrl":"https://doi.org/10.1109/BRACIS.2018.00083","url":null,"abstract":"In the CTO problem a group of moving observers should monitor a group of moving targets in order to maximize the average number of observed targets (ANOT). This work proposes and evaluates a centralized strategy based on FCM (Fuzzy C-means) to determine the movement of the observers in a setting in which the targets move randomly. The observer group is modeled as an agent organization specified in an FSB (Function-Structure-Behavior) standard. Compared with similar works, the test results are competitive in addition to avoiding known drawbacks reported in the literature.","PeriodicalId":405190,"journal":{"name":"2018 7th Brazilian Conference on Intelligent Systems (BRACIS)","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133253306","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/BRACIS.2018.00018
George Vilarinho, E. Ruiz
This paper presents a word graph based method for Twitter sentiment analysis (TSA) using global centrality metrics over graphs to evaluate sentiments, either positive or negative, expressed in microblogs. The proposed technique measures the importance of a sentence s for a given sentiment graph G by calculating its SentiElection coefficient. SentiElection, the method introduced in this work, is an ensemble of three global centrality measures: Katz index, Eigenvector centrality and PageRank. The results are compared to a previous model based on Containment similarity and Maximum Common Subgraph-based similarity metrics specifically designed to identify sentiments expressed in short texts. Using the geometric mean of their accuracies, we show the new suggested method outperforms the previous one.
{"title":"Global Centrality Measures in Word Graphs for Twitter Sentiment Analysis","authors":"George Vilarinho, E. Ruiz","doi":"10.1109/BRACIS.2018.00018","DOIUrl":"https://doi.org/10.1109/BRACIS.2018.00018","url":null,"abstract":"This paper presents a word graph based method for Twitter sentiment analysis (TSA) using global centrality metrics over graphs to evaluate sentiments, either positive or negative, expressed in microblogs. The proposed technique measures the importance of a sentence s for a given sentiment graph G by calculating its SentiElection coefficient. SentiElection, the method introduced in this work, is an ensemble of three global centrality measures: Katz index, Eigenvector centrality and PageRank. The results are compared to a previous model based on Containment similarity and Maximum Common Subgraph-based similarity metrics specifically designed to identify sentiments expressed in short texts. Using the geometric mean of their accuracies, we show the new suggested method outperforms the previous one.","PeriodicalId":405190,"journal":{"name":"2018 7th Brazilian Conference on Intelligent Systems (BRACIS)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133327376","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/BRACIS.2018.00078
A. Rivolli, C. Soares, A. Carvalho
In multi-label classification tasks, instances are simultaneously associated with multiple labels, representing different and, possibly, related concepts from a domain. One characteristic of these tasks is a high class-label imbalance. In order to obtain improved predictive models, several algorithms either have explored the label dependencies or have dealt with the problem of imbalanced labels. This work proposes a label expansion approach which combines both alternatives. For such, some labels are expanded with data from a related class label, making the labels more balanced and representative. Preliminary experiments show the effectiveness of this approach to improve the Binary Relevance strategy. Particularly, it reduced the number of labels that were never predicted in the test instances. Although the results are preliminary, they are potentially attractive, considering the scale and consistency of the improvement obtained, as well as the broad scope of the proposed approach.
{"title":"Label Expansion for Multi-label Classification","authors":"A. Rivolli, C. Soares, A. Carvalho","doi":"10.1109/BRACIS.2018.00078","DOIUrl":"https://doi.org/10.1109/BRACIS.2018.00078","url":null,"abstract":"In multi-label classification tasks, instances are simultaneously associated with multiple labels, representing different and, possibly, related concepts from a domain. One characteristic of these tasks is a high class-label imbalance. In order to obtain improved predictive models, several algorithms either have explored the label dependencies or have dealt with the problem of imbalanced labels. This work proposes a label expansion approach which combines both alternatives. For such, some labels are expanded with data from a related class label, making the labels more balanced and representative. Preliminary experiments show the effectiveness of this approach to improve the Binary Relevance strategy. Particularly, it reduced the number of labels that were never predicted in the test instances. Although the results are preliminary, they are potentially attractive, considering the scale and consistency of the improvement obtained, as well as the broad scope of the proposed approach.","PeriodicalId":405190,"journal":{"name":"2018 7th Brazilian Conference on Intelligent Systems (BRACIS)","volume":"2012 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129070466","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/BRACIS.2018.00070
Marlo Souza, Álvaro Freitas Moreira, R. Vieira
While several BDI logics have been proposed in the area of Agent Programming, it is not clear how these logics are connected to the agent programs they are supposed to specify. More yet, the reasoning problems in these logics, being based on modal logic, are not tractable in general, limiting their usage to tackle real-world problems. In this work, we use of Dynamic Preference Logic to provide a semantic foundation to BDI agent programming languages and investigate tractable expressive fragments of this logic to reason about agent programs. With that, we aim to provide a way of implementing semantically grounded agent programming languages with tractable reasoning cycles.
{"title":"Tractable Reasoning about Agent Programming in Dynamic Preference Logic","authors":"Marlo Souza, Álvaro Freitas Moreira, R. Vieira","doi":"10.1109/BRACIS.2018.00070","DOIUrl":"https://doi.org/10.1109/BRACIS.2018.00070","url":null,"abstract":"While several BDI logics have been proposed in the area of Agent Programming, it is not clear how these logics are connected to the agent programs they are supposed to specify. More yet, the reasoning problems in these logics, being based on modal logic, are not tractable in general, limiting their usage to tackle real-world problems. In this work, we use of Dynamic Preference Logic to provide a semantic foundation to BDI agent programming languages and investigate tractable expressive fragments of this logic to reason about agent programs. With that, we aim to provide a way of implementing semantically grounded agent programming languages with tractable reasoning cycles.","PeriodicalId":405190,"journal":{"name":"2018 7th Brazilian Conference on Intelligent Systems (BRACIS)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116504256","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：