POS Information Research Articles

A Transformer and Visual Foundation Model-Based Method for Cross-View Remote Sensing Image Retrieval

Abstract. Retrieving UAV images that lack POS information with georeferenced satellite orthoimagery is challenging due to the differences in angles of views. Most existing methods rely on deep neural networks with a large number of parameters, leading to substantial time and financial investments in network training. Consequently, these methods may not be well-suited for downstream tasks that have high timeliness requirements. In this work, we propose a cross-view remote sensing image retrieval method based on transformer and visual foundation model. We investigated the potential of visual foundation model for extracting common features from cross-view images. Training is only conducted on a small, self-designed retrieval head, alleviating the burden of network training. Specifically, we designed a CVV module to optimize the features extracted from the visual foundation model, making these features more adept for cross-view image retrieval tasks. And we designed an MLP head to achieve similarity discrimination. The method is verified on a publicly available dataset containing multiple scenes. Our method shows excellent results in terms of both efficiency and accuracy on 15 sub-datasets (10 or 50 scene categories) derived from the public dataset, which holds practical value in engineering applications with streamlined scene categories and constrained computational resources. Furthermore, we initiated a comprehensive discussion and conducted ablation experiments on the network design to validate its efficacy. Additionally, we analyzed the presence of overfitting within the network and deliberated on the limitations of our study, proposing potential avenues for future enhancements.

Assessing the consistency and reliability of the line intercept transect method in coral cover estimation using structure from motion photogrammetry techniques

Appropriate coral reef monitoring methods and descriptors determine the effectiveness of ecosystem status assessment. We combined line intercept transect (LIT) and SfM technologies for image acquisition, and the POS information of images is defined and assigned in the relative coordinate system we build. Then, constructed high-resolution orthomosaics of coral reefs along measuring tapes in Metashape. We compared the estimates of coral cover by three methodologies (LIT, LIT on orthomosaic (LITO), surface analyses on orthomosaic (SAO)) and used virtual transects on orthomosaics and their estimates of coral cover to assess the consistency and reliability of LIT. The results show that the estimation of coral cover by LITO or LIT is more consistent and reliable when coral cover is higher, and the probability of overestimation is lower. SfM technology is increasingly used in coral reef survey and assessment, and has the potential to become the new standard for coral reef survey.

Part-of-Speech Tags Guide Low-Resource Machine Translation

Neural machine translation models are guided by loss function to select source sentence features and generate results close to human annotation. When the data resources are abundant, neural machine translation models can focus on the features used to produce high-quality translations. These features include POS or other grammatical features. However, models cannot focus precisely on these features when data resources are limited. The reason is that the lack of samples makes the model overfit before considering these features. Previous works have enriched the features by integrating source POS or multitask methods. However, these methods only utilize the source POS or produce translations by introducing the generated target POS. We propose introducing POS information based on multitask methods and reconstructors. We obtain the POS tags by the additional encoder and decoder and compute the corresponding loss function. These loss functions are used with the loss function of machine translation to optimize the parameters of the entire model, which makes the model pay attention to POS features. The POS features focused on by models will guide the translation process and alleviate the problem that models cannot focus on the POS features in the case of low resources. Experiments on multiple translation tasks show that the method improves 0.4∼1 BLEU compared with the baseline model on different translation tasks.

The Construction of a Large-Scale Hong Kong Chinese Lexicon with Multilingual Translations for Chinese-as-an-Additional-Language (CAL) Students

This paper describes the structure and compilation process of a Hong Kong Chinese Lexicon being developed for a forthcoming Cantonese keyboard. The lexicon contains more than 120,000 headwords, including common words and high-frequency named entities, and incorporates information such as Jyutping romanization, indicative word frequency, simple translations in five languages (English, Hindi, Indonesian, Nepali, Urdu), disambiguation, POS information, speech register and cross-register synonyms. The lexicon improves the Chinese learning experience of CAL students in two ways: (a) introducing and promoting the effective usage of input methods and dictionaries for learners that facilitate them to locate the correct choice of word or phrase within a particular context, and (b) providing additional register-specific data that enable advanced learners to further expand their vocabulary. This lexicon can be used in conjunction with WordNet or LIWC (Linguistic Inquiry and Word Count) for a range of natural language processing tasks.

Fraud Resistant Off-Line Card Micropayments Using FRoDO and PUF

The internet payment system, which is a widespread cybercrime, is a major issue nowadays. Attackers concentrate on point of sale (POS) systems, the point at which a bank or merchant initially obtains customer information, with the goal of stealing confidential customer information. Effective computer systems with a card reader and specialised software are POS architectures. In this situation, the attacker might utilise malicious software (MS) to steal card statistics. If the customer and the supplier are consistently cut off from the network in, no online payment is practical. If the community fails, attacker's side will attempt to steal the password from the users throughout the price, therefore a safe online transaction price may not be achievable. Due to the PoS information breaches, we recommend an offline micro-fee solution that is secure and protects privacy for the persistent attackers in our paper. We use the FRoDO protocol to make the payment simple and secure from attackers. This protocol not only checks the client's coins but also confirms the client's identity by using identification details. This protocol increases flexibility and security and boosts the device's efficiency by infusing a comfortable micro-charge between the client and carrier.

Strategies in tracing linguistic variation in a corpus of Old Irish texts (CorPH)

AbstractThis article introduces Corpus PalaeoHibernicum (CorPH), a corpus currently consisting of 78 texts in Early Irish (c. 7th–10th cent.) created by the ERC-fundedChronologicon Hibernicum(ChronHib) project by bringing together pre-existing lexical and syntactic databases and adding further crucial texts from the period. In addition to being annotated for POS, morphological and syntactic information, another layer of annotation has been developed for CorPH – ‘Variation Tagging’, i.e. a tagset that numerically encodes synchronic language variation during the Early Irish period, thus allowing for much improved research on the chronological variation among the material. Another new pillar of studying linguistic variation is Bayesian Language Variation Analysis (BLaVA), in order to address the challenge that “not-so-big data” poses to statistical corpus methods. Instead of reflecting feature frequencies, BLaVA models language variation as probabilities of variation.

A Feature-Rich Vietnamese Named Entity Recognition Model

In this paper, we present a feature-based named entity recognition (NER) model that achieves the start-of-the-art accuracy for Vietnamese language. We combine word, word-shape features, PoS, chunk, Brown-cluster-based features, and word-embedding-based features in the Conditional Random Fields (CRF) model. We also explore the effects of word segmentation, PoS tagging, and chunking results of many popular Vietnamese NLP toolkits on the accuracy of the proposed feature-based NER model. Up to now, our work is the first work that systematically performs an extrinsic evaluation of basic Vietnamese NLP toolkits on the downstream NER task. Experimental results show that while automatically-generated word segmentation is useful, PoS and chunking information generated by Vietnamese NLP tools does not show their benefits for the proposed feature-based NER model.

Supervised Urdu Word Segmentation Model Based on POS Information

Supervised Urdu Word Segmentation Model Based on POS Information

Detecting proper nouns in indonesian-language translation of the quran using a guided method

Proper nouns (often abbreviated PN or NNP) are a class of words important in labelling and subsequent text processing, especially in natural language processing (NLP). Name entity recognition (NER) is one study that requires PN. The lack of labeled data for Indonesian text, especially the PN label, may be attributed for the lack of NER research in the Indonesian language. This study aims to detect PN in Indonesian-language translations of the Quran guided by deriving location information from the Quran as its source text. In the Indonesian language, PN are written using initial capital letters, which are used to determine and guide PN location. This article proposes that PN in Indonesian-language translations of the Quran can be determined based on PoS Information of Quranic text by developing a certain production rule. The results of this research showed that the proposed method has promising results for further research.

A Named Entity Recognition System for Malayalam using Neural Networks

Named Entity Recognition is the process of identifying the entities in the text document and categorizing them into predefined categories such as Person, Location, Organisation, etc. It is an important step in the processing of natural language text. Named entity recognition systems aim at extracting relevant information from the text. Various methods are applied for NER in Malayalam. In this paper, we propose an NER system for Malayalam using neural networks. Neural networks are exceptionally powerful tools for learning representations of data with multiple levels of abstraction. The proposed system utilizes different features such as POS information of the word, embedded representation of words and suffixes, POS information of preceding words, etc. We have used a corpus of 20615 sentences for training and testing. With less number of features, the system was able to obtain the state of the art performance in NER for Malayalam.

오피니언 마이닝을 위한 VOC 데이타의 신뢰성 분석

이 연구의 목적은 소셜 미디어에서 추출된 7개의 감성 도메인이 기업의 성과에 대한 영향 분석실험을 위한 데이터로서 적합한 지에 대해 신뢰성을 확인하고, 실제 고객감성이 자동차 시장점유율에 어떠한 영향을 미치는 지에 대하여 확인하기 위한 것이다. 본 연구는 총 3단계 구성으로서, 단계 1은 감성사전 구축 단계로서 미국 내 26개의 자동차 제조 회사의 고객의 소리 (VOC: Voice of Customer) 총 45,447개를 자동차 커뮤니티로부터 crawling하여 POS 정보 추출 후 감성사전을 구축하였고, 7개의 감성도메인을 만들었다. 단계 2는 신뢰성분석의 단계로서 자기상관관계분석과 주성분 분석 (PCA)을 통해 데이터의 실험 적합성을 검증하였다. 단계 3에서는 PCA를 근거로 2개의 선형회귀분석 모델을 구축하였고 GM, FCA, VOLKSWAGEN 등 3개의 기업을 선정, 2013년부터 2015년까지 7개 감성영역의 자동차 시장점유율에 대한 영향을 실험하였다. 실험 결과, 자기상관관계분석에 의해서 감성 데이터에 자기상관성과 시계열적 패턴이 관찰되었다. PCA 결과, 감성영역이 부정성, 긍정성, 중립성을 주성분으로 연결되어 있음이 확인되었다. VOC 감성 데이터에 대한 신뢰성을 바탕으로 한 2개 Model의 선형회귀분석 결과, 기업마다 시장점유율에 유의미한 영향을 미치는 감성들이 존재하며 Model 1과, 2의 감성영향력이 차이가 있고 중립성의 영향을 발견하였다. 본 연구를 통해, 데이터 상에 나타난 정보를 가진 감성이 과거 값에 기초하여 자동차 시장에서 변화를 수반할 수 있다는 것을 나타내고 있음을 확인하였다. 또한, 우리가 시장 데이터의 가용성을 적용하려고 할 때, 자동차 시장 관련 정보나 감성의 자기상관성을 잘 활용할 수 있다면, 감정 분석에 대한 연구에 큰 기여를 할 수 있을 뿐만 아니라, 실제 시장에서의 비지니스 성과에도 다양한 방법으로 기여할 수 있을 것으로 기대된다. The purpose of this study is to verify how 7 sentiment domains extracted through sentiment analysis from social media have an influence on business performance. It consists of three phases. In phase I, we constructed the sentiment lexicon after crawling 45,447 pieces of VOC (Voice of the Customer) on 26 auto companies from the car community and extracting the POS information and built a seven-sensitive domains. In phase II, in order to retain the reliability of experimental data, we examined auto-correlation analysis and PCA. In phase III, we investigated how 7 domains impact on the market share of three major (GM, FCA, and VOLKSWAGEN) auto companies by using linear regression analysis. The findings from the auto-correlation analysis proved auto-correlation and the sequence of the sentiments, and the results from PCA reported the 7 sentiments connected with positivity, negativity and neutrality. As a result of linear regression analysis on model 1, we indentified that the sentimental factors have a significant influence on the actual market share. In particular, not only posotive and negative sentiment domains, but neutral sentiment had significantly impacted on auto market share. As we apply the availability of data to the market, and take advantage of auto-correlation of the market-related information and the sentiment, the findings will be a huge contribution to other researches on sentiment analysis as well as actual business performances in various ways.

A POS Tagger for Social Media Texts Trained on Web Comments

Using social media tools such as blogs and forums have become more and more popular in recent years. Hence, a huge collection of social media texts from different communities is available for accessing user opinions, e.g., for marketing studies or acceptance research. Typically, methods from Natural Language Processing are applied to social media texts to automatically recognize user opinions. A fundamental component of the linguistic pipeline in Natural Language Processing is Part-of-Speech tagging. Most state-of-the-art Part-of-Speech taggers are trained on newspaper corpora, which differ in many ways from non-standardized social media text. Hence, applying common taggers to such texts results in performance degradation. In this paper, we present extensions to a basic Markov model tagger for the annotation of social media texts. Considering the German standard Stuttgart/T¨ ubinger TagSet (STTS), we distinguish 54 tag classes. Applying our approach improves the tagging accuracy for social media texts considerably, when we train our model on a combination of annotated texts from newspapers and Web comments. standardized text, since they are characterized by a spoken language, a dialogic and an informal writing style. This poses some special challenges to deal with in developing methods for automatic POS tagging of Web comments. These are particularly, the treatment of unknown (out-of-vocabulary) words and the different grammatical structure of social media texts in contrast to newspaper text. Furthermore, text genre specific manually annotated corpora, i.e., Web comments are required for training and testing. To the best of our knowledge all large manually annotated corpora are exclusively newspaper texts. In this work, we propose a Markov model tagger with parameter estimation enhancements for the POS annotation of social media texts. We apply and evaluate the tagger for German social media texts exemplarily. In order to make our method usable for NLP methods requiring POS information, e.g., syntactical parsing, we use the 54 Stuttgart/T¨ ubinger

Analysing trust as a means of improving the effectiveness of the virtual supply chain

This article focuses on trust within interorganisational virtual organisations, which are regarded as legally separate but operationally interdependent companies focused on responding to a market opportunity and facilitated by ICT as a medium for communication and coordination. The context of the research described in this paper is the Australian meat and horticulture supply chains. Within the context of these two supply chains the perceived importance and perceived levels of 12 trust factors are investigated. The research identifies three trust factors demonstrating critical gaps between the expected performance level and the perceptions of actual performance common to the meat and horticulture supply chains. These factors were information sharing, reliability and work standard. Two other critical factors were identified in the meat supply chain, i.e., timeliness and customisation, while three other critical factors were identified as specifically relating to the horticulture supply chain, i.e., shared values, POS information and honesty and integrity.

A Method for English-Korean Target Word Selection Using Multiple Knowledge Sources

Target word selection is one of the most important and difficult tasks in English-Korean Machine Translation. It effects on the overall translation accuracy of machine translation systems. In this paper, we present a new approach to Korean target word selection for an English noun with translation ambiguities using multiple knowledge such as verb frame patterns, sense vectors based on collocations, statistical Korean local context information and co-occurring POS information. Verb frame patterns constructed with dictionary and corpus play an important role in resolving the sparseness problem of collocation data. Sense vectors are a set of collocation data when an English word having target selection ambiguities is to be translated to specific Korean target word. Statistical Korean Local Context Information is an N-gram information generated using Korean corpus. The co-occurring POS information is a statistically significant POS clue which appears with ambiguous word. To evaluate our approach, we applied the method to Tellus-EK* system, English-Korean automatic translation system currently developed at ETRI [1],[2]. The experiment showed promising results for diverse sentences from web documents.

Integration of multiple evidences based on a query type for web search

The massive and heterogeneous Web exacerbates IR problems and short user queries make them worse. The contents of web pages are not enough to find answer pages. PageRank compensates for the insufficiencies of content information. The content information and PageRank are combined to get better results. However, static combination of multiple evidences may lower the retrieval performance. We have to use different strategies to meet the need of a user. We can classify user queries as three categories according to users' intent, the topic relevance task, the homepage finding task, and the service finding task. In this paper, we present a user query classification method. The difference of distribution, mutual information, the usage rate as anchor texts and the POS information are used for the classification. After we classified a user query, we apply different algorithms and information for the better results. For the topic relevance task, we emphasize the content information, on the other hand, for the homepage finding task, we emphasize the Link information and the URL information. We could get the best performance when our proposed classification method with the OKAPI scoring algorithm was used.

Mining inter-organizational retailing knowledge for an alliance formed by competitive firms

This paper applies data mining techniques to extract retailing knowledge from the POS information provided by an inter-organizational information service center in Taiwan. Many mutually competitive retail chains sponsored the data warehouse. They must, of course, protect their secrets, while cooperating to mine the inter-organizational data and thereby extract macro-level knowledge about consumers’ behavior. Many difficulties arise from this, because each transaction contains only a summary indicating the total sales of a single product in a store during a month and more detailed data are not available. Moreover, with many retail store chains cooperating, the meaning of the quantitative data, such as price and quantity, is difficult to compare and hard to interpret. No previous research addressed this problem. A series of steps were implemented to help solve this problem; they include defining semantic association rules (AR), transforming the quantitative data into semantic data and developing algorithms for mining the knowledge. Finally, we consolidated these ideas and implemented a prototype system.

Forecasting dynamic market share relationships

In market share analysis, it is fully recognized that we have often inadmissibly predicted market share, which means that some of predictors take the values outside the range [0, 1] and the total sum of predicted shares is not always one, so‐called “logical inconsistency”. Based on the Bayesian VAR model, proposes a dynamic market share model with logical consistency. The proposed method makes it possible to forecast not only the values of market share themselves, but also various dynamic market share relations across different brands or companies. The daily scanner data from the Nikkei POS information system are analyzed by the proposed method.

The Environmental Texture of Seafood Distribution in Japan

A startegic management approach is used to assess the nature of business environments affecting seafood distribution in Japan. Notions of environment ‘textures’ described in an influential paper by Emery and Trist are used to determine the current degree of environmental turbulence and the implications for export strategies. The environment as a whole is found to exhibit ‘medium turbulence’, and the sectors of importing, wholesaling, and food retailing/service are ‘medium’ to ‘high’ turbulence'. To cope with such environments, firms must undertake extensive real‐time market research (e.g. using POS information accessed via computer); develop flexible structures and operating systems (to become more responsive); and focus on improving communication and meeting the changing customer needs.