Iti Chaturvedi

Sentic LDA: Improving on LDA with semantic similarity for aspect-based sentiment analysis

The advent of the Social Web has provided netizens with new tools for creating and sharing, in a time- and cost-efficient way, their contents, ideas, and opinions with virtually the millions of people connected to the World Wide Web. This huge amount of information, however, is mainly unstructured as specifically produced for human consumption and, hence, it is not directly machine-processable. In order to enable a more efficient passage from unstructured information to structured data, aspect-based opinion mining models the relations between opinion targets contained in a document and the polarity values associated with these. Because aspects are often implicit, however, spotting them and calculating their respective polarity is an extremely difficult task, which is closer to natural language understanding rather than natural language processing. To this end, Sentic LDA exploits common-sense reasoning to shift LDA clustering from a syntactic to a semantic level. Rather than looking at word co-occurrence frequencies, Sentic LDA leverages on the semantics associated with words and multi-word expressions to improve clustering and, hence, outperform state-of-the-art techniques for aspect extraction.

The CLSA Model: A Novel Framework for Concept-Level Sentiment Analysis

Hitherto, sentiment analysis has been mainly based on algorithms relying on the textual representation of online reviews and microblogging posts. Such algorithms are very good at retrieving texts, splitting them into parts, checking the spelling, and counting their words. But when it comes to interpreting sentences and extracting opinionated information, their capabilities are known to be very limited. Current approaches to sentiment analysis are mainly based on supervised techniques relying on manually labeled samples, such as movie or product reviews, where the overall positive or negative attitude was explicitly indicated. However, opinions do not occur only at document-level, nor they are limited to a single valence or target. Contrary or complementary attitudes toward the same topic or multiple topics can be present across the span of a review. In order to overcome this and many other issues related to sentiment analysis, we propose a novel framework, termed concept-level sentiment analysis (CLSA) model, which takes into account all the natural-language-processing tasks necessary for extracting opinionated information from text, namely: microtext analysis, semantic parsing, subjectivity detection, anaphora resolution, sarcasm detection, topic spotting, aspect extraction, and polarity detection.

A report on single exon genes (SEG) in eukaryotes.

Single exon genes (SEG) are archetypical of prokaryotes. Hence, their presence in intron-rich, multi-cellular eukaryotic genomes is perplexing. Consequently, a study on SEG origin and evolution is important. Towards this goal, we took the first initiative of identifying and counting SEG in nine completely sequenced eukaryotic organisms--four of which are unicellular (E. cuniculi, S. cerevisiae, S. pombe, P. falciparum) and five of which are multi-cellular (C. elegans, A. thaliana, D. melanogaster, M. musculus, H. sapiens). This exercise enabled us to compare their proportion in unicellular and multi-cellular genomes. The comparison suggests that the SEG fraction decreases with gene count (r = -0.80) and increases with gene density (r = 0.88) in these genomes. We also examined the distribution patterns of their protein lengths in different genomes.

Bayesian network based extreme learning machine for subjectivity detection

Abstract Subjectivity detection is a task of natural language processing that aims to remove ‘factual’ or ‘neutral’ content, i.e., objective text that does not contain any opinion, from online product reviews. Such a pre-processing step is crucial to increase the accuracy of sentiment analysis systems, as these are usually optimized for the binary classification task of distinguishing between positive and negative content. In this paper, we extend the extreme learning machine (ELM) paradigm to a novel framework that exploits the features of both Bayesian networks and fuzzy recurrent neural networks to perform subjectivity detection. In particular, Bayesian networks are used to build a network of connections among the hidden neurons of the conventional ELM configuration in order to capture dependencies in high-dimensional data. Next, a fuzzy recurrent neural network inherits the overall structure generated by the Bayesian networks to model temporal features in the predictor. Experimental results confirmed the ability of the proposed framework to deal with standard subjectivity detection problems and also proved its capacity to address portability across languages in translation tasks.

Deep transfer learning for classification of time-delayed Gaussian networks

In this paper, we propose deep transfer learning for classification of Gaussian networks with time-delayed regulations. To ensure robust signaling, most real world problems from related domains have inherent alternate pathways that can be learned incrementally from a stable form of the baseline. In this paper, we leverage on this characteristic to address the challenges of complexity and scalability. The key idea is to learn high dimensional network motifs from low dimensional forms through a process of transfer learning. In contrast to previous work, we facilitate positive transfer by introducing a triangular inequality constraint, which provides a measure for the feasibility of mapping between different motif manifolds. Network motifs from different classes of Gaussian networks are used collectively to pre-train a deep neural network governed by a Lyapunov stability condition. The proposed framework is validated on time series data sampled from synthetic Gaussian networks and applied to a real world dataset for the classification of basketball games based on skill level. We observe an improvement in the range of 15-25% in accuracy and a saving in the range of 25-600% in computational cost on synthetic as well as realistic networks with time-delays when compared to existing state-of-the-art approaches. In addition, new insights into meaningful offensive formations in the Basketball games can be derived from the deep network. HighlightsIntroduce the deep transfer neural network (DTNN) which uses transfer learning to classify Gaussian networks with time delays.Transfer network motif ML probabilities from source to target network in order to save on computational cost associated with learning Gaussian networks.Determine feasibility of transfer from source to target manifolds via a triangular inequality condition.Encourage positive transfer by using network motifs with ML probability above a threshold to pre-train the DTNN.Learning of variable-order delays governed by a Lyapunov stability condition.Due to data sharing in the layers of the DTNN we require much fewer time samples compared to baselines.

Lyapunov filtering of objectivity for Spanish sentiment model

Objective sentences lack sentiments and, hence, can reduce the accuracy of a sentiment classifier. Traditional methods prior to 2001 used hand-crafted templates to identify subjectivity and did not generalize well for resource-deficient languages such as Spanish. Later works published between 2002 and 2009 proposed the use of deep neural networks to automatically learn a dictionary of features (in the form of convolution kernels) that is portable to new languages. Recently, recurrent neural networks are being used to model alternating subjective and objective sentences within a single review. Such networks are difficult to train for a large vocabulary of words due to the problem of vanishing gradients. Hence, in this paper we consider use of a Lyapunov linear matrix inequality to classify Spanish text as subjective or objective by combining Spanish features and features obtained from the corresponding translated English text. The aligned features for each sentence are next evolved using multiple kernel learning. The proposed Lyapunov deep neural network outperforms baselines by over 10% and the features learned in the hidden layers improve our understanding subjective sentences in Spanish.

Building gene networks with time-delayed regulations

We propose a method to build gene regulatory networks (GRN) capable of representing time-delayed regulations. The gene expression data is represented in two types of graphical models: a linear model using a dynamic Bayesian network (DBN) and a skip model using a hidden Markov model. The linear model is designed to find short-delays and skip model for long-delays. The algorithm was tested on time-series data obtained on yeast cell-cycle and validated against protein-protein interaction data. The proposed method better fits expression profiles compared to classical higher-order DBN and found core genes that are crucial in cell-cycle regulation.

Integration of epigenetic data in Bayesian network modeling of gene regulatory network

The reverse engineering of gene regulatory network (GRN) is an important problem in systems biology. While gene expression data provide a main source of insights, other types of data are needed to elucidate the structure and dynamics of gene regulation. Epigenetic data (e.g., histone modification) show promise to provide more insights into gene regulation and on epigenetic implication in biological pathways. In this paper, we investigate how epigenetic data are incorporated into reconstruction of GRN. We encode the histone modification data as prior for Bayesian network inference of GRN. Bayesian framework provides a natural and mathematically tractable way of integrating various data and knowledge through its prior. Applying to the gene expression data of yeast cell cycle, we demonstrate that integration of epigenetic data improves the accuracy of GRN inference significantly. Furthermore, fusion of gene expression and epigenetic data shed light on the interactions between genetic and epigenetic regulations of gene expression.

Gene regulatory networks with variable-order dynamic Bayesian networks

We introduce a probabilistic framework for building higher-order gene regulatory networks, which automatically finds the delays of regulatory interactions. A variable-order Markov chain Monte Carlo method with a new acceptance mechanism is proposed to find the optimal order and the structure of a dynamic Bayesian network (DBN). Experiments on cell cycle expression data indicate that the variable-order DBN (VDBN) better fits the data and gives biologically more plausible regulatory networks.

Detecting robust time-delayed regulation in Mycobacterium tuberculosis

BackgroundTime delays are often found in gene regulation though most techniques of building gene regulatory networks are not capable of capturing such phenomena. Here we look at the delays in the DNA repair system of Mycobacterium tuberculosis which is unusually slow in the bacteria. We propose a method based on a skip-chain model to study this phenomena in gene networks. The Viterbi paths of the underlying Markov chains find the most likely regulatory interactions among genes, taking care of very long delays. Using the derived networks, we discuss the delayed regulations and robustness of the DNA damage seen in the bacterium.ResultsWe evaluated our method on time-course gene expressions after DNA damage with Mitocyin C. Several time-delayed interactions were observed with our analysis. The presence of hubs in the networks indicates that a small number of transcriptional factors regulate the rest of the system. We demonstrate the use of priors to overcome over-fitting problem in the generation of networks. We compare our results with the gene networks derived with dynamic Bayesian networks (DBN).ConclusionDifferent transcription networks are active at different stages, and constant feedback and regulation is maintained throughout the activities of a biological pathway. Skip-chain models are capable of capturing, long distant and the time-delayed regulations. Use of a Dirichlet prior over parameters and Gibbs prior over structure can greatly reduce the over-fitting in the new model.