Efficient Vector Representation for Documents through Corruption

Balkan journal of electrical & computer engineering, 2024

The text classification task has a wide range of application domains for distinct purposes, such as the classification of articles, social media posts, and sentiments. As a natural language processing application, machine learning and deep learning techniques are intensively utilized in solving such challenges. One common approach is employing the discriminative word features comprising Bag-of-Words and n-grams to conduct text classification experiments. The other powerful approach is exploiting neural network-based (specifically deep learning models) through either sentence, word, or character levels. In this study, we proposed a novel approach to classify documents with contextually enriched word embeddings powered by the neighbor words accessible through the trigram word series. In the experiments, a well-known web of science dataset is exploited to demonstrate the novelty of the models. Consequently, we built various models constructed with and without the proposed approach to monitor the models' performances. The experimental models showed that the proposed neighborhood-based word embedding enrichment has decent potential to be used in further studies.

ArXiv, 2020

Efficient representation of text documents is an important building block in many NLP tasks. Research on long text categorization has shown that simple weighted averaging of word vectors for sentence representation often outperforms more sophisticated neural models. Recently proposed Sparse Composite Document Vector (SCDV) (Mekala et. al, 2017) extends this approach from sentences to documents using soft clustering over word vectors. However, SCDV disregards the multi-sense nature of words, and it also suffers from the curse of higher dimensionality. In this work, we address these shortcomings and propose SCDV-MS. SCDV-MS utilizes multi-sense word embeddings and learns a lower dimensional manifold. Through extensive experiments on multiple real-world datasets, we show that SCDV-MS embeddings outperform previous state-of-the-art embeddings on multi-class and multi-label text categorization tasks. Furthermore, SCDV-MS embeddings are more efficient than SCDV in terms of time and space ...

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics: Student Research Workshop

In document-level sentiment classification, each document must be mapped to a fixed length vector. Document embedding models map each document to a dense, lowdimensional vector in continuous vector space. This paper proposes training document embeddings using cosine similarity instead of dot product. Experiments on the IMDB dataset show that accuracy is improved when using cosine similarity compared to using dot product, while using feature combination with Naïve Bayes weighted bag of n-grams achieves a new state of the art accuracy of 97.42%.

2019

This work investigates the role of factors like training method, training corpus size and thematic relevance of texts in the performance of word embedding features on sentiment analysis of tweets, song lyrics, movie reviews and item reviews. We also explore specific training or post-processing methods that can be used to enhance the performance of word embeddings in certain tasks or domains. Our empirical observations indicate that models trained with multithematic texts that are large and rich in vocabulary are the best in answering syntactic and semantic word analogy questions. We further observe that influence of thematic relevance is stronger on movie and phone reviews, but weaker on tweets and lyrics. These two later domains are more sensitive to corpus size and training method, with Glove outperforming Word2vec. “Injecting” extra intelligence from lexicons or generating sentiment specific word embeddings are two prominent alternatives for increasing performance of word embedding features.

Springer, NLDB 2017 22nd International Conference on Natural Language & Information Systems, Liege, Belgium, 2017

Word embeddings or distributed representations of words are being used in various applications like machine translation, sentiment analysis, topic identification etc. Quality of word embeddings and performance of their applications depends on several factors like training method, corpus size and relevance etc. In this study we compare performance of a dozen of pretrained word embedding models on lyrics sentiment analysis and movie review polarity tasks. According to our results, Twitter Tweets is the best on lyrics sentiment analysis, whereas Google News and Common Crawl are the top performers on movie polarity analysis. Glove trained models slightly outrun those trained with Skipgram. Also, factors like topic relevance and size of corpus significantly impact the quality of the models. When medium or large-sized text sets are available, obtaining word embeddings from same training dataset is usually the best choice.

Journal of ICT Standardization

Natural Language Processing problems generally require the use of pre-trained distributed word representations to be solved with deep learning models. However, distributed representations usually rely on contextual information which prevents them from learning all the important word characteristics. The task of sentiment analysis suffers from such a problem because sentiment information is ignored during the process of learning word embeddings. The performance of sentiment analysis can be affected since two words with similar vectors may have opposite sentiment orientations. The present paper introduces a novel model called Continuous Sentiment Contextualized Vectors (CSCV) to address this problem. The proposed model can learn word sentiment embedding using its surrounding context words. It uses Continuous Bag-of-Words (CBOW) model to deal with the context and sentiment lexicons to identify sentiment. Existing pre-trained vectors are combined then with the obtained sentiment vectors...

arXiv (Cornell University), 2023

Learning quality document embeddings is a fundamental problem in natural language processing (NLP), information retrieval (IR), recommendation systems, and search engines. Despite recent advances in the development of transformer-based models that produce sentence embeddings with self-contrastive learning, the encoding of long documents (Ks of words) is still challenging with respect to both efficiency and quality considerations. Therefore, we train Longfomer-based document encoders using a state-of-the-art unsupervised contrastive learning method (SimCSE). Further on, we complement the baseline methodsiamese neural network-with additional convex neural networks based on functional Bregman divergence aiming to enhance the quality of the output document representations. We show that overall the combination of a self-contrastive siamese network and our proposed neural Bregman network outperforms the baselines in two linear classification settings on three long document topic classification tasks from the legal and biomedical domains.

Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval

Retrieving relevant documents from a corpus is typically based on the semantic similarity between the document content and query text. The inclusion of structural relationship between documents can benefit the retrieval mechanism by addressing semantic gaps. However, incorporating these relationships requires tractable mechanisms that balance structure with semantics and take advantage of the prevalent pre-train/fine-tune paradigm. We propose here a holistic approach to learning document representations by integrating intra-document content with inter-document relations. Our deep metric learning solution analyzes the complex neighborhood structure in the relationship network to efficiently sample similar/dissimilar document pairs and defines a novel quintuplet loss function that simultaneously encourages document pairs that are semantically relevant to be closer and structurally unrelated to be far apart in the representation space. Furthermore, the separation margins between the documents are varied flexibly to encode the heterogeneity in relationship strengths. The model is fully finetunable and natively supports query projection during inference. We demonstrate that it outperforms competing methods on multiple datasets for document retrieval tasks. CCS CONCEPTS • Information systems → Retrieval models and ranking.

We propose two novel model architectures for computing continuous vector representations of words from very large data sets. The quality of these representations is measured in a word similarity task, and the results are compared to the previously best performing techniques based on different types of neural networks. We observe large improvements in accuracy at much lower computational cost, i.e. it takes less than a day to learn high quality word vectors from a 1.6 billion words data set. Furthermore, we show that these vectors provide state-of-the-art performance on our test set for measuring syntactic and semantic word similarities. arXiv:1301.3781v3 [cs.CL] 7 Sep 2013 1 The test set is available at www.fit.vutbr.cz/˜imikolov/rnnlm/word-test.v1.txt 2

The bag-of-words (BOW) model is the common approach for classifying documents, where words are used as feature for training a classifier. This generally involves a huge number of features. Some techniques, such as Latent Semantic Analysis (LSA) or Latent Dirichlet Allocation (LDA), have been designed to summarize documents in a lower dimension with the least semantic information loss. Some semantic information is nevertheless always lost, since only words are considered. Instead, we aim at using information coming from n-grams to overcome this limitation, while remaining in a low-dimension space. Many approaches, such as the Skip-gram model, provide good word vector representations very quickly. We propose to average these representations to obtain representations of n-grams. All n-grams are thus embedded in a same semantic space. A K-means clustering can then group them into semantic concepts. The number of features is therefore dramatically reduced and documents can be represented as bag of semantic concepts. We show that this model outperforms LSA and LDA on a sentiment classification task, and yields similar results than a traditional BOW-model with far less features.