Abstract

These days' discussion forums provide dependable solutions to the problems related to multiple domains and areas. However, due to the presence of huge amount of less-informative/inappropriate posts, the identification of the appropriate problem-solution pairs has become a challenging task. The emergence of a variety of topics, domains and areas has made the task of manual labelling of the problem solution-post pairs a very costly and time consuming task. To solve these issues, we concentrate on deep semantic and logical relation between terms. For this, we introduce a novel semantic correlation graph to represent the text. The proposed representation helps us in the identification of topical and semantic relation between terms at a fine grain level. Next, we apply the improved version of personalised pagerank using random walk with restarts. The main aim is to improve the rank score of terms having direct or indirect relation with terms in the given question. Finally, we introduce the use of the node overlapping version of GAAC to find the actual span of answer text. Our experimental results show that the devised system performs better than the existing unsupervised systems.

Abstract

In this paper, we present a novel N-gram (N> = 1) filtration technique for keyphrase extraction. To filter the sophisticated candidate keyphrases (N-grams), we introduce the combined use of: 1) statistical feature (obtained by using weighted betweenness centrality scores of words, which is generally used to identify the border nodes/edges in community detection techniques); 2) co-location strength (calculated by using nearest neighbour Dbpedia texts). We also introduce the use of N-gram (N> = 1) graph, which reduces the bias effect of lower length N-grams in the ranking process and preserves the semantics of words (phraseness), based upon local context. To capture the theme of the document and to reduce the effect of noisy terms in the ranking process, we apply an information theoretic framework for key-player detection on the proposed N-gram graph. Our experimental results show that the devised system performs better than the current state-of-the-art unsupervised systems and comparable/better than supervised systems.

Abstract

An overwhelming amount of data is generated everyday on social media, encompassing a wide spectrum of topics. With almost every business decision depending on customer opinion, mining of social media data needs to be quick and easy. For a data analyst to keep up with the agility and the scale of the data, it is impossible to bank on fully supervised techniques to mine topics and their associated sentiments from social media. Motivated by this, we propose a weakly supervised approach (named, TweetGrep) that lets the data analyst easily define a topic by few keywords and adapt a generic sentiment classifier to the topic–by jointly modeling topics and sentiments using label regularization. Experiments with diverse datasets show that TweetGrep beats the state-ofthe-art models for both the tasks of retrieving topical tweets and analyzing the sentiment of the tweets (average improvement of 4.97% and 6.91% respectively in terms of area under the curve). Further, we show that TweetGrep can also be adopted in a novel task of hashtag disambiguation, which significantly outperforms the baseline methods.

Abstract

In this paper, we consider the problem of learning representations for authors from bibliographic co-authorship networks. Existing methods for deep learning on graphs, such as DeepWalk, suffer from link sparsity problem as they focus on modeling the link information only. We hypothesize that capturing both the content and link information in a unified way will help mitigate the sparsity problem. To this end, we present a novel model'Author2Vec', which learns low-dimensional author representations such that authors who write similar content and share similar network structure are closer in vector space. Such embeddings are useful in a variety of applications such as link prediction, node classification, recommendation and visualization. The author embeddings we learn are empirically shown to outperform DeepWalk by 2.35% and 0.83% for link prediction and clustering task respectively.

Abstract

Extractive summarization techniques typically aim to maximize the information coverage of the summary with respect to the original corpus and report accuracies in ROUGE scores. Automated text summarization techniques should consider the dimensions of comprehensibility, coherence and readability. In the current work, we identify the discourse structure which provides the context for the creation of a sentence. We leverage the information from the structure to frame a monotone (non-decreasing) sub-modular scoring function for generating comprehensible summaries. Our approach improves the overall quality of comprehensibility of the summary in terms of human evaluation and gives sufficient content coverage with comparable ROUGE score. We also formulate a metric to measure summary comprehensibility in terms of Contextual Independence of a sentence. The metric is shown to be representative of human judgement of text comprehensibility.

Abstract

Doc2Sent2Vec is an unsupervised approach to learn lowdimensional feature vector (or embedding) for a document. This embedding captures the semantics of the document and can be fed as input to machine learning algorithms to solve a myriad number of applications in the field of data mining and information retrieval. Some of these applications include document classification, retrieval, and ranking. The proposed approach is two-phased. In the first phase, the model learns a vector for each sentence in the document using a standard word-level language model. In the next phase, it learns the document representation from the sentence sequence using a novel sentence-level language model. Intuitively, the first phase captures the word-level coherence to learn sentence embeddings, while the second phase captures the sentence-level coherence to learn document embeddings. Compared to the state-of-the-art models that learn document vectors directly from the word sequences, we hypothesize that the proposed decoupled strategy of learning sentence embeddings followed by document embeddings helps the model learn accurate and rich document representations. We evaluate the learned document embeddings by considering two classification tasks: scientific article classification and Wikipedia page classification. Our model outperforms the current state-of-the-art models in the scientific article classification task by∼ 12.07% and the Wikipedia page classification task by∼ 6.93%, both in terms of F1 score. These results highlight the superior quality of document embeddings learned by the Doc2Sent2Vec approach

Abstract

With improving efficiency and cost effectiveness of public cloud systems, there has been a growing trend to complement in-house cloud environment with them. We propose an open solution to the problem of distributing jobs in a hybrid environment, as a substitute to hadoop systems which are cloud specific. The proposed system, MultiStack, is a big data orchestration platform for deploying big data jobs across multiple cloud providers. The specific architecture elaborated in this paper uses Amazon Web Services as the public cloud provider and Openstack as the private cloud framework. Our solution supports complete Hadoop ecosystem tools - hive, pig, Hbase, oozie etc. and on-demand scaling of Hadoop clusters. The proposed framework aims at reducing the Job completion time on workloads along with decrease in cost using Spot Instance provisioning compared to on-demand provisioning. This is achieved by providing two modes of operation: Proactive scheduling and Reactive scheduling, which takes into account user providing job characteristics(eg memory, cpu, etc), quota limitation and business objectives. From our experiments, we conclude that Multistack is able to reduce average job completion time by 30-37% with minimal increase in cost.

Abstract

The problem of providing optimal assignment for backend storage is a central problem in the design of cloud systems. It has taken a further central role as a result of growing heterogeneity from emerging Software Defined Storage systems. In this paper, we propose a solution to optimal IO Workload assignment using statistical modelling to estimate measures of performance such as Throughput, IOPS, et al. The proposed system uses support vector regression to estimate the performance of individual IO Workloads on each available SDS system for optimal assignment. As a proof of concept, we demonstrate our solution in a heterogeneous environment comprising of HDFS, GlusterFS, and Ceph. We first show the accuracy of estimation of throughput and IOPS with values of coefficient of determination over 0.65 in all cases. We further show the analysis of using this regression model to classify workloads to respective SDS backend that will maximize throughput.

Abstract

Research in social media analysis is experiencing a recent surge with a large number of works applying representation learning models to solve high-level syntactico-semantic tasks such as sentiment analysis, semantic textual similarity computation, hashtag prediction and so on. Although the performance of the representation learning models are better than the traditional baselines for the tasks, little is known about the core properties of a tweet encoded within the representations. Understanding these core properties would empower us in making generalizable conclusions about the quality of representations. Our work presented here constitutes the first step in opening the black-box of vector embedding for social media posts, with emphasis on tweets in particular. In order to understand the core properties encoded in a tweet representation, we evaluate the representations to estimate the extent to which it can model each of those properties such as tweet length, presence of words, hashtags, mentions, capitalization, and so on. This is done with the help of multiple classifiers which take the representation as input. Essentially, each classifier evaluates one of the syntactic or social properties which are arguably salient for a tweet. This is also the first holistic study on extensively analysing the ability to encode these properties for a wide variety of tweet representation models including the traditional unsupervised methods (BOW, LDA), unsupervised representation learning methods (Siamese CBOW, Tweet2Vec) as well as supervised methods (CNN, BLSTM).

Abstract

Sentiment analysis (SA) using code-mixed data from social media has several applications in opinion mining ranging from customer satisfaction to social campaign analysis in multilingual societies. Advances in this area are impeded by the lack of a suitable annotated dataset. We introduce a Hindi-English (Hi-En) code-mixed dataset for sentiment analysis and perform empirical analysis comparing the suitability and performance of various state-of-the-art SA methods in social media. In this paper, we introduce learning sub-word level representations in our LSTM (Subword-LSTM) architecture instead of character-level or word-level representations. This linguistic prior in our architecture enables us to learn the information about sentiment value of important morphemes. This also seems to work well in highly noisy text containing misspellings as shown in our experiments which is demonstrated in morpheme-level feature maps learned by our model. Also, we hypothesize that encoding this linguistic prior in the Subword-LSTM architecture leads to the superior performance. Our system attains accuracy 4-5% greater than traditional approaches on our dataset, and also outperforms the available system for sentiment analysis in Hi-En code-mixed text by 18%.