Search

Financial sentiment analysis is a very challenging problem because the market is influenced by various factors, such as company-specific/political news, sentiment/opinions of users, and other regional financial market. Good news can drive the market to grow positively, while negative news can drag the market downwards. For this reason, it is crucial to understand the impacts of news and social media on the stock market trends. Motivated by this, this paper focuses on developing an effective and efficient company-specific financial sentiment analysis model which can detect the trends of a company's stock price. More specifically, we develop a novel neural network model that transforms pretrained general word embeddings into domain-specific embeddings. In addition, we use a knowledge-base to enrich the training vocabulary, and thus extend the domain-specific embedding space. The main challenge for natural language processing (NLP) applications is to learn the representation for the rare and unseen words. Another challenge for financial sentiment analysis models addressed in this paper is to deal with words that change their polarities depending upon the domain in which they are used. We thoroughly evaluate the performance of the proposed model on a benchmark dataset of SemEval-2017 shared task on financial sentiment analysis. The experimental results show that the proposed model delivers state-of-the-art performance when applied on Twitter and news headlines datasets, thus demonstrating its feasibility and effectiveness. [ABSTRACT FROM AUTHOR]

Cross-Lingual Information Retrieval (CLIR) provides flexibility to users to query in their regional (source) languages regardless the target documents languages. CLIR uses trending translation techniques Statistical Machine Translation (SMT) and Neural Machine Translation (NMT). SMT and NMT achieve good results for foreign languages but not for Indian languages due to non-absoluteness of the parallel corpus. Source language user queries may contain the Out Of Vocabulary (OOV) words which are not present in the parallel corpus such words may be skipped without performing translation by SMT. In this paper, a context-based translation algorithm is proposed to translate the OOV words by utilizing two unlabeled & unrelated large raw corpora (in source and target language) and a small bi-lingual parallel corpus. Since SMT performs better than NMT for Hindi to English translation as per the literature, therefore, experimental results are evaluated for FIRE datasets against baseline SMT. The proposed algorithm improves evaluation measures, Recall up to 6.04% (0.8785) for FIRE 2010 and up to 3.96% (0.7365) for FIRE 2011, & Mean Average Precision (MAP) up to 14.37% (0.3239) for FIRE 2010 and up to 5.46% (0.1988) for FIRE 2011, in comparison to the baseline SMT which achieves 0.8284 and 0.7084 Recall for FIRE 2010 and 2011, & 0.2832 and 0.1885 MAP for FIRE 2010 and 2011. An analysis for the number of OOV words shows that the proposed algorithm reduces the number of OOV more effectively, up to 0.81% for FIRE 2010 and 1.73% for FIRE 2011. [ABSTRACT FROM AUTHOR]

Persian writing system deviates from the ideal one due to the lack of one-to-one correspondence between graphemes and phonemes. The present study deals with this question that in spite of the absence of short vowels in Persian writing system and one-to-many and many-to-one relationships between the graphemes and phonemes, how can Persian speakers read out of vocabulary words? This study introduces a probabilistic- structural method that Persian speakers use to read out of vocabulary words in which structural information (including Persian morphology and morphophonemic rules) as well as Arabic morphological templates are considered. In order to test how the introduced method works, Persian speakers were asked to read a list of out of vocabulary words. The mentioned list was used by ID3 and MLP (two methods which are used in machine learning) as input, then the outputs of the method and those of ID3 and MLP were compared with Persian speakers` pronunciations the results proved that the introduced method functions similar to Persian speakers in reading out of vocabulary words.

Copyright of Procesamiento del Lenguaje Natural is the property of Sociedad Espanola para el Procesamiento del Lenguaje Natural and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Persian writing system deviates from the ideal one due to the lack of one-to-one correspondence between graphemes and phonemes. The present study deals with this question that in spite of the absence of short vowels in Persian writing system and one-to-many and many-to-one relationships between the graphemes and phonemes, how can Persian speakers read out of vocabulary words? This study introduces a probabilistic- structural method that Persian speakers use to read out of vocabulary words in which structural information (including Persian morphology and morphophonemic rules) as well as Arabic morphological templates are considered. In order to test how the introduced method works, Persian speakers were asked to read a list of out of vocabulary words. The mentioned list was used by ID3 and MLP (two methods which are used in machine learning) as input, then the outputs of the method and those of ID3 and MLP were compared with Persian speakers` pronunciations the results proved that the introduced method functions similar to Persian speakers in reading out of vocabulary words. [ABSTRACT FROM AUTHOR]