Abstract Although deep neural networks are excellent for text sentiment analysis, their applications in real-world practice occasionally limited owing to black-box property. In this study, we propose a novel network model called contextual (CSNN) that can explain the process of its analysis prediction way humans find natural and agreeable catch up summary contents. The CSNN has following interpretable layers: word-level original layer, shift global importance concept-level layer. Because...

A linear multi-factor model is one of the most important tools in equity portfolio management. The models are widely used because they can be easily interpreted. However, financial markets not and their accuracy limited. Recently, deep learning methods were proposed to predict stock return terms model. Although these perform quite well, have significant disadvantages such as a lack transparency limitations interpretability prediction. It thus difficult for institutional investors use...

Word-level contextual sentiment analysis (WCSA) is an important task for mining reviews or opinions. When analyzing this type of in the industry, both interpretability and practicality are often required. However, such a WCSA method has not been established. This study aims to develop with practicality. To achieve aim, we propose novel neural network architecture called Sentiment Interpretable Neural Network (SINN). realize SINN practically, learning strategy Lexical Initialization Learning...

In this research, two estimation algorithms for extracting cross-lingual news pairs based on machine learning from financial articles have been proposed. Every second, innumerable text data, including all kinds news, reports, messages, reviews, comments, and tweets are generated the Internet, these written not only in English but also other languages such as Chinese, Japanese, French, etc. By taking advantage of multi-lingual resources provided by Thomson Reuters News, we developed...

Although deep neural networks are excellent for text sentiment analysis, their applications in real-world practice occasionally limited owing to black-box property. In response, we propose a novel network model called contextual (CSNN) that can explain the process of its analysis prediction way humans find natural and agreeable. The CSNN has following interpretable layers: word-level original layer, shift local global importance layer. Because these layers, this document-level results...

This paper proposes a data storage method that cost-efficiently manages massive amount of short lifetime continuously generated typically by IoT devices (called Live Data, e.g., surveillance camera images) considering real-time retrieval. Along with the rapid growth IoT, many users expect to shift an Open in which they can share enormous amounts Data. To efficiently use it is important both hot and cold storage. Data should be stored storage, less expensive but has slow access speeds than...

Extracting useful information for market trend analysis automatically from textual data is an important issue in the financial field. Textual Yahoo finance bulletin board posts one of valuable resources. From these posts, we can obtain such as investors' sentiments and discussions between investors. On boards, number untagged more than ten times that with sentiment tags. Thus, believe indicators extracted both unlabeled labeled represent better those only posts. However, no other previous...

Bidders often take a long time to read and understand tender documents because they require specialized knowledge, are generally long. first overview the specific items, such as payment warranty, in document then check overall document. Therefore, function that can extract items (i.e., item extractor) highlight words or phrases related word-phrase highlighter) great demand. To develop above two types of functions, we need solve problems. The problem is annotated data set. second concerns...

In Information-Centric Network or location-free contents network, it is a big challenge to manage metadata of the data continuously generated with short lifetime (called Live Data), for example, surveillance camera data. This paper proposes new architecture suitable handling Data by considering its feature high locality in retrieval. The includes methods (1) having local databases and categorizing into global (2) assigning MAC address storage content ID metadata. Quantitative evaluations...

This paper presents the performance evaluation of a metadata database (DB) management method that uses realistic numeric examples for IoT Live Data. The is proposed to reduce handling costs Data are here defined as data typically continuously generated by devices and have short lifetimes (e.g., 10 fps surveillance camera images). We already an model in which high locality significantly featured usage. previous results obtained only from general parameter values statistical distributions. To...

The purpose of our research is to estimate positive-negative scores new financial terms, and find the feature vector a document useful for stock price prediction. We propose technology calculate word's score from existing words' scores. First, we assigned numerical word appeared in news documents using word2vec algorithm, defined vectors documents. Then, analyzed trends sentiment which can be evaluated textual data Yahoo! finance board these vectors. As result comparison with other...

Automatic creation of polarity dictionaries is an important issue, as explanations prediction models are often required in the financial industry. This paper proposes a novel method developing interpretable and predictable neural network model. The model we built can extract scores concepts from documents. Furthermore, detect pairwise interactions between concepts, create concept using our was vector representations words for about 100 provided by professionals, obtained hundred times more...

This paper proposes a new Information-Centric Network architecture (ICN) suitable for handling metadata of the data which is continuously generated typically by IoT devices and/or has short lifetime (called Live Data). The proposed ICN two methods considering feature high locality Data retrieval; (1) stored queries either on databases (DB) or generation terminals (publishers) with popularity and data, query sent to DB publishers in local area issue (subscribers). (2) DBs consist one global...

Zero-rating service provided by Mobile Virtual Network Operators (MVNOs) has been attracting smartphone users who frequently watch web videos that are delivered heavily bandwidth-consuming applications. With the increase of encrypted traffic, MVNOs need to identify video hosting sites accessed via traffic analysis for enabling such services. If from permitted is identified as coming non-permitted due mistaken identification sites, unreasonable payments inevitable or subscribers, and vice...

It has a great demand for automatically visualizing word-level sentiment scores in financial documents the form that even non-experts can briefly understand documents. In this paper, we aim to develop method original (i.e., before considering contexts document) and contextual after of each term document. To achieve aim, assigning both words using Layer-wise Relevance Propagation (LRP) method. The LRP based approach consider information sentiments words, contrast other approaches. Using...