Semantic Tags Research Articles

Metaphor in second language academic writing

This study describes the quantitative aspects of metaphor in academic publications. In particular, we analyze the frequency of metaphor types and source domains in medical papers written by native and non-native speakers of English. The data have been collected from selected published articles on cancer and tissue engineering. The method combines a manual procedure with a semantic tagger and keyness statistics in Wmatrix software for metaphor identification. The results show that while the distribution of metaphor is relatively stable across tissue engineering and cancer articles written by native and non-native speakers, indirect metaphor is more often used than other types of metaphor. Moreover, the findings indicate that some metaphorical source domains are specifically used for a particular topic and that some are more frequently exploited by native speakers of English than by non-native speakers. This study has implications for English for Academic Purposes (EAP) pedagogy and the development of metaphorical awareness activities as it may set the ground for future corpus studies that analyze metaphor properties in academic registers.

Dual-Level Semantic Transfer Deep Hashing for Efficient Social Image Retrieval

Social network stores and disseminates a tremendous amount of user shared images. Deep hashing is an efficient indexing technique to support large-scale social image retrieval, due to its deep representation capability, fast retrieval speed and low storage cost. Particularly, unsupervised deep hashing has well scalability as it does not require any manually labelled data for training. However, owing to the lacking of label guidance, existing methods suffer from severe semantic shortage when optimizing a large amount of deep neural network parameters. Differently, in this paper, we propose a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Dual-level Semantic Transfer Deep Hashing</i> (DSTDH) method to alleviate this problem with a unified deep hash learning framework. Our model targets at learning the semantically enhanced deep hash codes by specially exploiting the user-generated tags associated with the social images. Specifically, we design a complementary dual-level semantic transfer mechanism to efficiently discover the potential semantics of tags and seamlessly transfer them into binary hash codes. On the one hand, instance-level semantics are directly preserved into hash codes from the associated tags with adverse noise removing. Besides, an image-concept hypergraph is constructed for indirectly transferring the latent high-order semantic correlations of images and tags into hash codes. Moreover, the hash codes are obtained simultaneously with the deep representation learning by the discrete hash optimization strategy. Extensive experiments on two public social image retrieval datasets validate the superior performance of our method compared with state-of-the-art hashing methods. The source codes of our method can be obtained at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/research2020-1/DSTDH</uri>

Quality Assessment of Semantic Tags in OpenStreetMap

A comprehensive understanding of urban areas includes deep knowledge about locations and functions of buildings within. In most developed countries cadastral data is available for this research, but from a global perspective the only free and comprehensive data source is OpenStreetMap (OSM).For this study we selected 42 cities across the globe covering a wide range of climate zones as well as cultures and assess the accuracy of building function labels in OSM indirectly by comparing them to Google Places. We state that points-of-interest (POIs) are reasonably covered in Google Places due to a large number of users and a business perspective driving its development. We study how many semantic building tags are in accordance with the proposed scheme by OSM. In this regard Los Angeles has the best coverage, followed by Amsterdam and Cologne. In Melbourne, Paris, and Sydney we find the most matching building functions of OSM and Google Places.In summary, we conclude that OSM is not ready to provide ground truth labels for each place on the globe, but can serve as a powerful and rich label source in selected study areas. Our study gives a first insight where obtaining training labels from OSM is a valid and reliable approach.

Digging Deeper into the Finnish Parliamentary Protocols

This paper analyses the protocols of the Parliament of Finland 1907–2000. They have been digitised and published as open data by the Finnish Parliament in 20181. In the analysis we use a novel tool, a semantic tagger for Finnish – FiST [1]. We describe the tagger generally and show results of seman-tic analysis both on the whole of the parliamentary corpus and on a small subset of data where everyman’s rights (a widely used right of public access to nature) have been the main topic of parliamentary discussions. Our analysis contributes to the understanding of the development of this “tradition” of public access rights, and is also the first study utilizing the Finnish semantic tagger as a tool for content analysis in digital humanities research. Keyword search shows first that that the discussion of everyman’s rights has had three different peak peri-ods in the Finnish Parliament: 1946, 1973, and 1992. Secondly, the contents of the discussions have different nature for all the periods, which could be clearly detected with FiST and keyness analysis.

Architecture to improve the accuracy of automatic image annotation systems

Automatic image annotation (AIA) is an image retrieval mechanism to extract relative semantic tags from visual content. So far, the improvement of accuracy in newly developed such methods have been about 1 or 2% in the F1‐score and the architectures seem to have room for improvement. Therefore, the authors designed a more detailed architecture for AIA and suggested new algorithms for its main parts. The proposed architecture has three main parts: feature extraction, learning, and annotation. They designed a novel learning method using machine learning and probability bases. In the annotation part, they suggest a novel method that gains the maximum benefit from the learning part. The combination of the proposed architecture, algorithms, and novel ideas resulted in new accuracy milestones in F1‐score on most commonly used datasets. In their architecture, N+ measure which shows the number of tags with non‐zero recalls showed that they could recall all tags for IAPRTC‐12 and ESP‐Games datasets.

Why the fruit picker smiles in an anti-corruption story: Analyzing evaluative clash and news value construction in online news discourse in China

In China’s journalism, positive photographs are preferred in reporting on negative news stories and such divergent couplings of text and image are crucial in newsworthiness mediation in helping readers understand social values in China. Drawing on Discursive News Value Analysis and appraisal, this article investigates the construal of Negativity and Positivity through evaluative resources across visual-verbal semiotic modes in China’s English-language news discourse. Sourced from China Daily and People’s Daily online websites, 34 news reports featuring evaluative clash are studied by combining corpus semantic tagging method and manual analysis from both the text-as-corpus and text-as-text approaches. The findings show that Negativity is largely and strategically construed through negative lexis and evaluative language towards news participants like China, the US and Australia in international disputes, while visual Positivity plays an important role in orienting readers to the Socialist Core Value of “Harmony” against the negative news stories through their interplay in the heading-image nucleus. Underlying journalistic positions are revealed through close analysis of news value co-construction in three news stories about national issues, in which visual devices are employed to enhance, contradict or complement the linguistic expressions around social value building.

SEMANTIC GROUP IN THE LANGUAGE CORPUS, THE IMPORTANCE OF SEMANTIC TEXT SETTING OF GROUP AND FIELD EXPLANATION

There are morphological, syntactic, semantic, accentic aspects of layout in the language corpus. The corpus is the most convenient means of observing the language fluctuation of the dictionary (neologism, istorism, archaism) and its material must be linguistically processed. The complex of tags is the main tool for the implementation of this linguistic process (analysis). By monitoring the surroundings of this or that linguistic unit, it is possible to identify certain semantic characters that characterize this unity. The word may refer to a few semantic categories simultaneously, so it is important to consider the level of the word or the category of the word. The rank is determined by counting the distribution frequency according to different categories. Thus, by means of a corpus it is possible to find out the semantic category of words and to get acquainted with the meaning of each category. A set of semantic tags should be perfectly designed so that the body can indicate which semantic category it belongs to. This article discusses the significance of semantic groups, colonies and domains in the development of a set of semantic labels. The set of semantic tags is based on the semantic field, semantic colon, and LSG. On the first page of the interface semantic fields of the Uzbek language are given, if a semantic field is selected, the second window opens, the semantic groups of the same area appear, semantic gangs are again separated from the LSG. This structure of the interface works on the basis of the analysis of “analyze the word” principle. At the stage of morphological markup for the horse horoscope, the semantic area affiliation also refers to an explanation of how the LMG enters the label. The use of the semantic theory of the body in the case of a corpus can also be the basis for the ideological arc of the body.

Understanding Smartphone Users From Installed App Lists Using Boolean Matrix Factorization.

Smartphones are changing humans' lifestyles. Mobile applications (apps) on smartphones serve as entries for users to access a wide range of services in our daily lives. The apps installed on one's smartphone convey lots of personal information, such as demographics, interests, and needs. This provides a new lens to understand smartphone users. However, it is difficult to compactly characterize a user with his/her installed app list. In this article, a user representation framework is proposed, where we model the underlying relations between apps and users with Boolean matrix factorization (BMF). It builds a compact user subspace by discovering basic components from installed app lists. Each basic component encapsulates a semantic interpretation of a series of special-purpose apps, which is a reflection of user needs and interests. Each user is represented by a linear combination of the semantic basic components. With this user representation framework, we use supervised and unsupervised learning methods to understand users, including mining user attributes, discovering user groups, and labeling semantic tags to users. Extensive experiments were conducted on three data subsets from a large-scale real-world dataset for evaluation, each consisting of installed app lists from over 10 000 users. The results demonstrated the effectiveness of our user representation framework.

Multiple instance deep learning for weakly-supervised visual object tracking

Intelligently tracking objects with varied shapes, color, lighting conditions, and backgrounds is an extremely useful application in many HCI applications, such as human body motion capture, hand gesture recognition, and virtual reality (VR) games. However, accurately tracking different objects under uncontrolled environments is a tough challenge due to the possibly dynamic object parts, varied lighting conditions, and sophisticated backgrounds. In this work, we propose a novel semantically-aware object tracking framework, wherein the key is weakly-supervised learning paradigm that optimally transfers the video-level semantic tags into various regions. More specifically, give a set of training video clips, each of which is associated with multiple video-level semantic tags, we first propose a weakly-supervised learning algorithm to transfer the semantic tags into various video regions. The key is a MIL (Zhong et al., 2020) [1]-based manifold embedding algorithm that maps the entire video regions into a semantic space, wherein the video-level semantic tags are well encoded. Afterward, for each video region, we use the semantic feature combined with the appearance feature as its representation. We designed a multi-view learning algorithm to optimally fuse the above two types of features. Based on the fused feature, we learn a probabilistic Gaussian mixture model to predict the target probability of each candidate window, where the window with the maximal probability is output as the tracking result. Comprehensive comparative results on a challenging pedestrian tracking task as well as the human hand gesture recognition have demonstrated the effectiveness of our method. Moreover, visualized tracking results have shown that non-rigid objects with moderate occlusions can be well localized by our method.

Providing effective recommendations in discussion groups using a new hybrid recommender system based on implicit ratings and semantic similarity

Abstract Discussion groups are one of the most important elements of collaborative learning which utilize recommender systems to improve their performance in several aspects. This type of learning facilitates a comfort communication between users to share their problems and questions and receive the appropriate solutions. Most of recommender systems of discussion groups are based on using collaborative filtering techniques and a few numbers of them use content-based or hybrid filtering. Experimental results of previous works show that using hybrid recommender systems on discussion groups’ databases cause significant improvement in accuracy of recommended posts in comparison with other filtering techniques (Kardan and Ebrahimi, 2013). To improve performance of (Kardan and Ebrahimi, 2013), in this paper, a new recommender system is represented, which includes three parts, namely content-based, collaborative, and hybrid filtering parts. The proposed recommender system uses the tagging features to provide more appropriate recommendations on discussion groups. For this purpose, semantic relevance of tags is extracted using WordNet lexical database and the tags are organized in a hierarchical structure based on their semantic relevance. The hierarchical structure is used for searching relevant posts in content-based filtering part, and the user’s query is extended using related semantic tags. The implicit ratings of the users are calculated in the collaborative filtering part using similarity measures. Finally, the results of these two parts are combined in the hybrid filtering part of the proposed system to recommend the posts of the discussion group which are similar to the query of the active user. Experimental results show higher precision of the proposed system comparing to the former recommender systems.

A semantic assessment framework for e-learning systems

Semantic web technologies are applied in a wide spectrum of applications including search engines, web services discovery and composition, semantic tags and electronic assessment of e-learning systems. Recently, the field of electronic assessment has grabbed the attention of many researchers. Electronic assessment is a real challenge taking to consideration the diversity of question types. E-assessment of answers to open questions has created new challenges regarding extracting students' answers using natural language processing techniques and infer new knowledge using description logic reasoning. To be able to automate the process of assessing the students' answers, they have to be annotated with semantics. This paper presents a thorough survey of significant research efforts that adopt semantic annotation to enhance the process of e-assessment. Moreover, in this paper, we propose a semantic-based framework that automates the process of electronically assessing the answers of questions on e-learning systems based on their semantic descriptions.

A semantic assessment framework for e-learning systems

Semantic web technologies are applied in a wide spectrum of applications including search engines, web services discovery and composition, semantic tags and electronic assessment of e-learning systems. Recently, the field of electronic assessment has grabbed the attention of many researchers. Electronic assessment is a real challenge taking to consideration the diversity of question types. E-assessment of answers to open questions has created new challenges regarding extracting students' answers using natural language processing techniques and infer new knowledge using description logic reasoning. To be able to automate the process of assessing the students' answers, they have to be annotated with semantics. This paper presents a thorough survey of significant research efforts that adopt semantic annotation to enhance the process of e-assessment. Moreover, in this paper, we propose a semantic-based framework that automates the process of electronically assessing the answers of questions on e-learning systems based on their semantic descriptions.

An Efficient and Effective Model Based on Mean Positive Examples for Social Image Annotation

Nowadays, with the rapid growth of imaging and social network, huge volumes of image data are produced and shared on social media. Social image annotation has been an important and challenging task in the fields of computer vision and machine learning, which can facilitate large-scale image retrieval, indexing, and management. The four most challenges of social image annotation are semantic gap, tag refinement, label-imbalance, and annotation efficiency. To address these issues, we propose an efficient and effective annotation method based on the Mean of Positive Examples (MPE) corresponding to each label. First, we refine user-provided noisy tags by our proposed local smoothing process, and consider the refined tags as key features in contrast to the previous methods that consider them as side information, which significantly improves annotation performance. Second, we propose a weighted trans-media similarity measure method that fuses all modality information in identifying proper neighbors, which promotes the semantic level and eases image annotation. Third, our MPE model gives equal importance to all labels, thus, improving the annotation performance of infrequent labels without sacrificing that of frequent labels. Fourth, our MPE model can dramatically decrease space-time overheads, since the time cost of annotating an image is unaffected by the size of the training image dataset, but relying on the size of label vocabulary. Therefore, our proposed method can be applied to real-world large-scale online social image repositories. Extensive experiments on both benchmark datasets demonstrate the effectiveness and efficiency of our MPE model.

CLASSIFICATION OF SHORT POSSESSIVE CLITIC PRONOUN NYA IN MALAY TEXT TO SUPPORT ANAPHOR CANDIDATE DETERMINATION

Anaphor candidate determination is an important process in anaphora resolution (AR) systems. There are several types of anaphor, one of which is pronominal anaphor. Pronominal anaphor is an anaphor that involves pronouns. In some of the cases, certain pronouns can be used without referring to any situation or entity in a text, and this phenomenon is known as pleonastic. In the case of the Malay language, it usually occurs for the pronoun nya. The pleonastic that exists in every text causes a severe problem to the anaphora resolution systems. The process to determine the pleonastic nya is not the same as identifying the pleonastic ‘it’ in the English language, where the syntactic pattern could not be used because the structure of nya comes at the end of a word. As an alternative, semantic classes are used to identify the pleonastic itself and the anaphoric nya. In this paper, the automatic semantic tag was used to determine the type of nya, which at the same time could determine nya as an anaphor candidate. The new algorithms and MalayAR architecture were proposed. The results of the F-measure showed the detection of clitic nya as a separate word achieved a perfect 100% result. In comparison, the clitic nya as a pleonastic achieved 88%, clitic nya referring to humans achieved 94%, and clitic nya referring to non-humans achieved 63%. The results showed that the proposed algorithms were acceptable to solve the issue of the clitic nya as pleonastic, human referral as well as non-human referral.

Hierarchical attentive deep neural networks for semantic music annotation through multiple music representations

Automatically assigning a group of appropriate semantic tags to one music piece provides an effective way for people to efficiently utilize the massive and ever increasing online and off-line music data. In this paper, we propose a novel end-to-end deep neural network model for automatic music annotation, which effectively integrates available multiple complementary music representations and jointly accomplishes music representation learning, structure modeling, and tag prediction. The model first hierarchically leverages attentive convolutional networks and recurrent networks to learn informative descriptions from Mel-spectrogram and raw waveform of the music and depict time-varying structures embedded in the description sequence. A dual-state LSTM network is then employed to capture the correlations between two representation channels as supplementary music descriptions. Finally, the model aggregates music description sequence into a holistic embedding with a self-attentive multi-weighting mechanism, which adaptively captures multi-aspect summarized information of the music for tag prediction. Experiments on the public MagnaTagATune benchmark music dataset show that the proposed model outperforms state-of-the-art methods for automatic music annotation.

Semantic Tagging of Singing Voices in Popular Music Recordings

Singing voice is a key sound source in popular music. As recent music streaming and entertainment services call for more intelligent solutions to retrieve songs or evaluate musical characteristics, automatic analysis of popular music targeted to singing voice has been a significant research subject. The majority of studies have focused on quantitative or objective information of singing voice such as pitch, lyrics or singer identity. However, singing voice has a wide variety of dimensions that are somewhat difficult to quantify and therefore we often describe by words. In this article, we address the qualitative analysis of singing voice as a music auto-tagging task that annotates songs with a set of tag words. To this end, we build a music tag dataset dedicated to singing voice. Specifically, we define a vocabulary that describes timbre and singing styles of K-pop vocalists and collect human annotations for individual tracks. We then conduct statistical analysis to understand the global and temporal characteristics of the tag words. Using the dataset, we train a deep neural network model to automatically predict the voice-specific tags from popular music recordings and evaluate the model in different conditions. We discuss the results by comparing them to the statistical analysis of tag words. Finally, we show potential applications of the vocal tagging system in music retrieval, music thumbnailing and singing evaluation.

A latent semantic analysis-based image tag optimisation method

According to the nominal tags, the adverb and adjective high-level semantic tags which is described human emotion are easily to exceed the handling scope. Furthermore, the majority of optimisation methods exists many problems (e.g., although inputting a high application cost, it cannot get a satisfactory effect). This paper proposed an effective image tag optimisation algorithm which was composed of two important parts: firstly, a random walk model was used to process the initial image tag information, the image relationship diagram and the tag relationship diagram were structured based on image vision similarity and tag relevancy respectively, then, the image and tag information was spread through a dual-diagram (image relationship diagram and tag relationship diagram) walk mode randomly; secondly, a new image tag optimisation model was built according to three factors, i.e., semantic uniformity, noise sparsity and result matrix sparsity. In the experimental stage, the effectiveness of this model and the superiority of pre-processing were verified by experiment. Compared to other methods, the experimental result indicates that this algorithm is more reasonable and efficient.

Computationally Analyzing Social Media Text for Topics: A Primer for Advertising Researchers

Advertising researchers and practitioners are increasingly using social media analytics (SMA), but focused overviews that explain how to use various SMA techniques are scarce. We focus on how researchers and practitioners can computationally analyze topics of conversation in social media posts, compare each to a human-coded topic analysis of a brand’s Twitter feed, and provide recommendations on how to assess and choose which computational methods to use. The computational methodologies that we survey in this article are text preprocessed summarization, phrase mining, topic modeling, supervised machine learning for text classification, and semantic topic tagging.

Learning Compact Spatio-Temporal Features for Fast Content based Video Retrieval

Videos are recorded and uploaded daily to the sites like YouTube, Facebook etc. from devices such as mobile phones and digital cameras with less or without metadata (semantic tags) associated with it. This makes extremely difficult to retrieve similar videos based on this metadata without using content based semantic search. Content based video retrieval is problem of retrieving most similar videos to a given query video and has wide range of applications such as video browsing, content filtering, video indexing, etc. Traditional video level features based on key frame level hand engineered features which does not exploit rich dynamics present in the video. In this paper we propose a fast content based video retrieval framework using compact spatio-temporal features learned by deep learning. Specifically, deep CNN along with LSTM is deploy to learn spatio-temporal representations of video. For fast retrieval, binary code is generated by hashing learning component in the framework. For fast and effective learning of hash code proposed framework is trained in two stages. First stage learns the video dynamics and in second stage compact code is learn using learned video’s temporal variation from the first stage. UCF101 dataset is used to test the proposed method and results compared by other hashing methods. Results show that our approach is able to improve the performance over existing methods.

Content Based Movie Scene Retrieval using Spatio Temporal Features

Thousands of movies along with TV shows, documentaries are being produced each year around the world with different genres and languages. Making a movie scene impactful as well as original is challenging task for the director. On the other hand, users demands to retrieve similar scenes from their queries is also challenging task as there is no proper maintenance of database of movie scene videos with proper semantic tags associated with it. So to fulfill the requirement of these two (but not the least) application areas there is a need of content based retrieval system for movie scenes. Content based video retrieval is a problem of retrieving most similar videos to a given query video by analyzing the visual contents of videos. Traditional video level features based on key frame level hand engineered features which does not exploit rich dynamics present in the video. In this paper we propose a Content based Movie Scene Retrieval (CB-MSR) framework using spatio-temporal features learned by deep learning. Specifically deep CNN along with LSTM is deploy to learn spatio-temporal representations of video. On the basis of these learned features similar movie scenes can be retrieve from the collection of movies. Hollywood2 dataset is used to test the proposed system. Two types of features: spatial and spatio-temporal features are used to evaluate the proposed framework.