Embedding Hierarchical Structures for Venue Category Representation

Venue categories used in location-based social networks often exhibit a hierarchical structure, together with the category sequences derived from users’ check-ins. The two data modalities provide a wealth of information for us to capture the semantic relationships between those categories. To understand the venue semantics, existing methods usually embed venue categories into low-dimensional spaces by modeling the linear context (i.e., the positional neighbors of the given category) in check-in sequences. However, the hierarchical structure of venue categories, which inherently encodes the relationships between categories, is largely untapped. In this article, we propose a venue C ategory E mbedding M odel named Hier-CEM , which generates a latent representation for each venue category by embedding the Hier archical structure of categories and utilizing multiple types of context. Specifically, we investigate two kinds of hierarchical context based on any given venue category hierarchy and show how to model them together with the linear context collaboratively. We apply Hier-CEM to three tasks on two real check-in datasets collected from Foursquare. Experimental results show that Hier-CEM is better at capturing both semantic and sequential information inherent in venues than state-of-the-art embedding methods.

TRG-DAtt: The Target Relational Graph and Double Attention Network Based Sentiment Analysis and Prediction for Supporting Decision Making

The management of public opinion and the use of big data monitoring to accurately judge and verify all kinds of information are valuable aspects in the enterprise management decision-making process. The sentiment analysis of reviews is a key decision-making tool for e-commerce development. Most existing review sentiment analysis methods involve sequential modeling but do not focus on the semantic relationships. However, Chinese semantics are different from English semantics in terms of the sentence structure. Irrelevant contextual words may be incorrectly identified as cues for sentiment prediction. The influence of the target words in reviews must be considered. Thus, this paper proposes the TRG-DAtt model for sentiment analysis based on target relational graph (TRG) and double attention network (DAtt) to analyze the emotional information to support decision making. First, dependency tree-based TRG is introduced to independently and fully mine the semantic relationships. We redefine and constrain the dependency and use it as the edges to connect the target and context words. Second, we design dependency graph attention network (DGAT) and interactive attention network (IAT) to form the DAtt and obtain the emotional features of the target words and reviews. DGAT models the dependency of the TRG by aggregating the semantic information. Next, the target emotional enhancement features obtained by the DGAT are input to the IAT. The influence of each target word on the review can be obtained through the interaction. Finally, the target emotional enhancement features are weighted by the impact factor to generate the review's emotional features. In this study, extensive experiments were conducted on the car and Meituan review data sets, which contain consumer reviews on cars and stores, respectively. The results demonstrate that the proposed model outperforms the existing models.

Self-Attention Networks and Adaptive Support Vector Machine for Aspect-Level Sentiment Classification

Aspect-level sentiment classification aims to integrating the context to predict the sentiment polarity of aspect-specific in a text, which has been quite useful and popular, e.g. opinion survey and products’ recommending in e-commerce. Many recent studies exploit a Long Short-Term Memory (LSTM) networks to perform aspect-level sentiment classification, but the limitation of long-term dependencies is not solved well, so that the semantic correlations between each two words of the text are ignored. In addition, traditional classification model adopts SoftMax function based on probability statistics as classifier, but ignores the words’ features in the semantic space. Support Vector Machine (SVM) can fully use the information of characteristics and it is appropriate to make classification in the high dimension space, however which just considers the maximum distance between different classes and ignores the similarities between different features of the same classes. To address these defects, we propose the two-stages novel architecture named Self Attention Networks and Adaptive SVM (SAN-ASVM) for aspect-level sentiment classification. In the first-stage, in order to overcome the long-term dependencies, Multi-Heads Self Attention (MHSA) mechanism is applied to extract the semantic relationships between each two words, furthermore 1-hop attention mechanism is designed to pay more attention on some important words related to aspect-specific. In the second-stage, ASVM is designed to substitute the SoftMax function to perform sentiment classification, which can effectively make multi-classifications in high dimensional space. Extensive experiments on SemEval2014, SemEval2016 and Twitter datasets are conducted, compared experiments prove that SAN-ASVM model can obtains better performance.

Completing WordNets with Sememe Knowledge

WordNets organize words into synonymous word sets, and the connections between words present the semantic relationships between them, which have become an indispensable source for natural language processing (NLP) tasks. With the development and evolution of languages, WordNets need to be constantly updated manually. To address the problem of inadequate word semantic knowledge of “new words”, this study explores a novel method to automatically update the WordNet knowledge base by incorporating word-embedding techniques with sememe knowledge from HowNet. The model first characterizes the relationships among words and sememes with a graph structure and jointly learns the embedding vectors of words and sememes; finally, it synthesizes word similarities to predict concepts (synonym sets) of new words. To examine the performance of the proposed model, a new dataset connected to sememe knowledge and WordNet is constructed. Experimental results show that the proposed model outperforms the existing baseline models.

Cross-modal Semantic Relationships Guide Spontaneous Orienting in Real-life Scenes

In real-world scenes, the different objects and events available to our senses are interconnected within a rich web of semantic associations. These semantic links help parse information and make sense of the environment. For example, during goal-directed attention, characteristic everyday life object sounds help speed up visual search for these objects in natural and dynamic environments. However, it is not known whether semantic correspondences also play a role under spontaneous observation. Here, we investigated this question addressing whether crossmodal semantic congruence can drive spontaneous, overt visual attention in free-viewing conditions. We used eye-tracking whilst participants (N=45) viewed video clips of realistic complex scenes presented alongside sounds of varying semantic congruency with objects within the videos. We found that characteristic sounds increased the probability of looking, the number of fixations, and the total dwell time on the semantically corresponding visual objects, in comparison to when the same scenes were presented with semantically neutral sounds or just with background noise only. Our results suggest that crossmodal semantic congruence has an impact on spontaneous gaze and eye movements, and therefore on how attention samples information in a free viewing paradigm. Our findings extend beyond known effects of object-based crossmodal interactions with simple stimuli and shed new light upon how audio-visual semantically congruent relationships play out in everyday life scenarios.

Matching Cyber Security Ontologies through Genetic Algorithm-Based Ontology Alignment Technique

Security ontology can be used to build a shared knowledge model for an application domain to overcome the data heterogeneity issue, but it suffers from its own heterogeneity issue. Finding identical entities in two ontologies, i.e., ontology alignment, is a solution. It is important to select an effective similarity measure (SM) to distinguish heterogeneous entities. However, due to the complex semantic relationships among concepts, no SM is ensured to be effective in all alignment tasks. The aggregation of SMs so that their advantages and disadvantages complement each other directly affects the quality of alignments. In this work, we formally define this problem, discuss its challenges, and present a problem-specific genetic algorithm (GA) to effectively address it. We experimentally test our approach on bibliographic tracks provided by OAEI and five pairs of security ontologies. The results show that GA can effectively address different heterogeneous ontology-alignment tasks and determine high-quality security ontology alignments.

If a picture is worth 1000 words, is a word worth 1000 features for design metric estimation?

A picture is worth a thousand words, and in design metric estimation, a word may be worth a thousand features. Pictures are awarded this worth because they encode a plethora of information. When evaluating designs, we aim to capture a range of information, including usefulness, uniqueness, and novelty of a design. The subjective nature of these concepts makes their evaluation difficult. Still, many attempts have been made and metrics developed to do so, because design evaluation is integral to the creation of novel solutions. The most common metrics used are the consensual assessment technique (CAT) and the Shah, Vargas-Hernandez, and Smith (SVS) method. While CAT is accurate and often regarded as the “gold standard,” it relies on using expert ratings, making CAT expensive and time-consuming. Comparatively, SVS is less resource-demanding, but often criticized as lacking sensitivity and accuracy. We utilize the complementary strengths of both methods through machine learning. This study investigates the potential of machine learning to predict expert creativity assessments from non-expert survey results. The SVS method results in a text-rich dataset about a design. We utilize these textual design representations and the deep semantic relationships that natural language encodes to predict more desirable design metrics, including CAT metrics. We demonstrate the ability of machine learning models to predict design metrics from the design itself and SVS survey information. We show that incorporating natural language processing improves prediction results across design metrics, and that clear distinctions in the predictability of certain metrics exist.

Folk Definitions as a Model for Sign Language Dictionary Definitions: A User-Focused Study of the Online Dictionary of New Zealand Sign Language

<p>This thesis addresses the question whether signed definitions, made possible by advances in electronic lexicography, should be introduced to sign language dictionaries. The thesis comprises four interrelated studies investigating different aspects of this question through a user-focused case study of the Online Dictionary of New Zealand Sign Language (ODNZSL). A preliminary study investigated current use of the ODNZSL in order to identify what user needs signed definitions might fulfil. The study drew on two data sets: website log data for the ODNZSL, and a think-aloud protocol and interview with representatives of user groups. Results showed that in addition to a large volume of casual browsers, the most frequent and intensive users of the dictionary are beginner and intermediate students of New Zealand Sign Language (NZSL). These (hearing) language learners mostly search for frequent vocabulary with the aims of language production and vocabulary learning. Findings also identified reasons for unsuccessful dictionary consultations that may impact on the effectiveness of definitions. In the second study, a review of ODNZSL entries highlighted categories of lexical items for which the current description through English glosses, examples, and usage notes is inadequate. A test was developed to assess whether these categories of signs were problematic for the user group identified in the first study: hearing intermediate learners of NZSL. Twenty-one participants took a computer-based error correction test with both comprehension and production sections comprising fifty items in six different categories: culture-bound; idiomatic; polysemous; metaphoric/metonymic; vocabulary type / word class; and other. Quantitative results indicated that a small number of test items were problematic, but that none of the test categories were good predictors of the difficulties learners experienced. A qualitative examination identified linguistic factors and issues with the current dictionary information that may be improved by the addition of signed definitions. The central proposition tested in the third study was that folk definitions—informal explanations of sign meaning by Deaf sign language users—can be applied as a template for dictionary definitions. This study took fifteen of the signs that were identified as problematic for learners in the previous study, and asked thirteen Deaf NZSL users to explain the meaning of these signs. A qualitative analysis found that the folk definitions by different NZSL users shared common semantic categories and embedded information about situational and sociolinguistic variation as well as grammatical structures. Some semantic relationships that occur frequently in spoken language folk definitions, such as exemplification and synonymy, were also common in signed folk definitions. Other semantic relationships such as attribution, function, operation, and spatial relationships occurred less frequently because they were inherent in the sign construction. Due to the bilingual status of the participants, many folk definitions included reference to English words in the form of mouth patterns and fingerspelling. In the fourth study, twelve pilot dictionary definitions were created on the basis of common features found in the folk definitions and an evaluation of definition formats by Deaf NZSL users. The error correction test from the second study was repeated with a new cohort of intermediate NZSL learners. This time twelve test items were accompanied by a pilot definition; for the remaining items participants were shown a video example sentence from the ODNZSL entry. Results showed no significant improvements in scores for the test items with definitions. However, feedback from test participants showed that the definitions were comprehensible and perceived as valuable for language learning. The overall conclusion of these studies is that a selective approach should be taken to introducing signed definitions in existing multifunctional sign language dictionaries. For hearing learners of sign language, signed definitions do not meet immediate communicative (comprehension and production) needs, but they may contribute to wider vocabulary learning goals. The main contribution of this thesis is that it suggests a user-focused methodology for creating signed definitions, driven by evidence from the first empirical user study of an online sign language dictionary and therefore taking into account the particular challenges of sign language lexicography. Furthermore, the analysis of features of signed folk definitions contributes to the semantic description of sign languages.</p>

Folk Definitions as a Model for Sign Language Dictionary Definitions: A User-Focused Study of the Online Dictionary of New Zealand Sign Language

<p>This thesis addresses the question whether signed definitions, made possible by advances in electronic lexicography, should be introduced to sign language dictionaries. The thesis comprises four interrelated studies investigating different aspects of this question through a user-focused case study of the Online Dictionary of New Zealand Sign Language (ODNZSL). A preliminary study investigated current use of the ODNZSL in order to identify what user needs signed definitions might fulfil. The study drew on two data sets: website log data for the ODNZSL, and a think-aloud protocol and interview with representatives of user groups. Results showed that in addition to a large volume of casual browsers, the most frequent and intensive users of the dictionary are beginner and intermediate students of New Zealand Sign Language (NZSL). These (hearing) language learners mostly search for frequent vocabulary with the aims of language production and vocabulary learning. Findings also identified reasons for unsuccessful dictionary consultations that may impact on the effectiveness of definitions. In the second study, a review of ODNZSL entries highlighted categories of lexical items for which the current description through English glosses, examples, and usage notes is inadequate. A test was developed to assess whether these categories of signs were problematic for the user group identified in the first study: hearing intermediate learners of NZSL. Twenty-one participants took a computer-based error correction test with both comprehension and production sections comprising fifty items in six different categories: culture-bound; idiomatic; polysemous; metaphoric/metonymic; vocabulary type / word class; and other. Quantitative results indicated that a small number of test items were problematic, but that none of the test categories were good predictors of the difficulties learners experienced. A qualitative examination identified linguistic factors and issues with the current dictionary information that may be improved by the addition of signed definitions. The central proposition tested in the third study was that folk definitions—informal explanations of sign meaning by Deaf sign language users—can be applied as a template for dictionary definitions. This study took fifteen of the signs that were identified as problematic for learners in the previous study, and asked thirteen Deaf NZSL users to explain the meaning of these signs. A qualitative analysis found that the folk definitions by different NZSL users shared common semantic categories and embedded information about situational and sociolinguistic variation as well as grammatical structures. Some semantic relationships that occur frequently in spoken language folk definitions, such as exemplification and synonymy, were also common in signed folk definitions. Other semantic relationships such as attribution, function, operation, and spatial relationships occurred less frequently because they were inherent in the sign construction. Due to the bilingual status of the participants, many folk definitions included reference to English words in the form of mouth patterns and fingerspelling. In the fourth study, twelve pilot dictionary definitions were created on the basis of common features found in the folk definitions and an evaluation of definition formats by Deaf NZSL users. The error correction test from the second study was repeated with a new cohort of intermediate NZSL learners. This time twelve test items were accompanied by a pilot definition; for the remaining items participants were shown a video example sentence from the ODNZSL entry. Results showed no significant improvements in scores for the test items with definitions. However, feedback from test participants showed that the definitions were comprehensible and perceived as valuable for language learning. The overall conclusion of these studies is that a selective approach should be taken to introducing signed definitions in existing multifunctional sign language dictionaries. For hearing learners of sign language, signed definitions do not meet immediate communicative (comprehension and production) needs, but they may contribute to wider vocabulary learning goals. The main contribution of this thesis is that it suggests a user-focused methodology for creating signed definitions, driven by evidence from the first empirical user study of an online sign language dictionary and therefore taking into account the particular challenges of sign language lexicography. Furthermore, the analysis of features of signed folk definitions contributes to the semantic description of sign languages.</p>

Context aware semantic adaptation network for cross domain implicit sentiment classification

Cross-domain sentiment classification could be attributed to two steps. The first step is used to extract the text representation, and the other is to reduce domain discrepancy. Existing methods mostly focus on learning the domain-invariant information, rarely consider using the domain-specific semantic information, which could help cross-domain sentiment classification; traditional adversarial-based models merely focus on aligning the global distribution ignore maximizing the class-specific decision boundaries. To solve these problems, we propose a context-aware semantic adaptation (CASA) network for cross-domain implicit sentiment classification (ISC). CASA can provide more semantic relationships and an accurate understanding of the emotion-changing process for ISC tasks lacking explicit emotion words. (1) To obtain inter- and intrasentence semantic associations, our model builds a context-aware heterogeneous graph (CAHG), which can aggregate the intrasentence dependency information and the intersentence node interaction information, followed by an attention mechanism that remains high-level domain-specific features. (2) Moreover, we conduct a new multigrain discriminator (MGD) to effectively reduce the interdomain distribution discrepancy and improve intradomain class discrimination. Experimental results demonstrate the effectiveness of different modules compared with existing models on the Chinese implicit emotion dataset and four public explicit datasets.