| 1 |
Front Matter |
|
|
Abstract
|
| 2 |
|
|
|
Abstract
|
| 3 |
Open-Set Web Genre Identification Using Distributional Features and Nearest Neighbors Distance Ratio |
Dimitrios Pritsos,Anderson Rocha,Efstathios Stamatatos |
|
Abstract
Web genre identification can boost information retrieval systems by providing rich descriptions of documents and enabling more specialized queries. The open-set scenario is more realistic for this task as web genres evolve over time and it is not feasible to define a universally agreed genre palette. In this work, we bring to bear a novel approach to web genre identification underpinned by distributional features acquired by doc2vec and a recently-proposed open-set classification algorithm—the nearest neighbors distance ratio classifier. We present experimental results using a benchmark corpus and a strong baseline and demonstrate that the proposed approach is highly competitive, especially when emphasis is given on precision.
|
| 4 |
Exploiting Global Impact Ordering for Higher Throughput in Selective Search |
Michał Siedlaczek,Juan Rodriguez,Torsten Suel |
|
Abstract
We investigate potential benefits of exploiting a global impact ordering in a selective search architecture. We propose a generalized, ordering-aware version of the learning-to-rank-resources framework [.] along with a modified selection strategy. By allowing partial shard processing we are able to achieve a better initial trade-off between query cost and precision than the current state of the art. Thus, our solution is suitable for increasing query throughput during periods of peak load or in low-resource systems.
|
| 5 |
Cross-Domain Recommendation via Deep Domain Adaptation |
Heishiro Kanagawa,Hayato Kobayashi,Nobuyuki Shimizu,Yukihiro Tagami,Taiji Suzuki |
|
Abstract
The behavior of users in certain services indicates their preferences, which may be used to make recommendations for other services they have never used. However, the cross-domain relation between items and user preferences is not simple, especially when there are few or no common users and items across domains. We propose a content-based cross-domain recommendation method for cold-start users that does not require user- or item-overlap. We formulate recommendations as an extreme classification task, and the problem is treated as an instance of unsupervised domain adaptation. We assess the performance of the approach in experiments on large datasets collected from Yahoo! JAPAN video and news services and find that it outperforms several baseline methods including a cross-domain collaborative filtering method.
|
| 6 |
It’s only Words and Words Are All I Have |
Manash Pratim Barman,Kavish Dahekar,Abhinav Anshuman,Amit Awekar |
|
Abstract
The central idea of this paper is to demonstrate the strength of lyrics for music mining and natural language processing (NLP) tasks using the distributed representation paradigm. For music mining, we address two prediction tasks for songs: genre and popularity. Existing works for both these problems have two major bottlenecks. First, they represent lyrics using handcrafted features that require intricate knowledge of language and music. Second, they consider lyrics as a weak indicator of genre and popularity. We overcome both the bottlenecks by representing lyrics using distributed representation. In our work, genre identification is a multi-class classification task whereas popularity prediction is a binary classification task. We achieve an F1 score of around 0.6 for both the tasks using only lyrics. Distributed representation of words is now heavily used for various NLP algorithms. We show that lyrics can be used to improve the quality of this representation.
|
| 7 |
Modeling User Return Time Using Inhomogeneous Poisson Process |
Mohammad Akbari,Alberto Cetoli,Stefano Bragaglia,Andrew D. O’Harney,Marc Sloan,Jun Wang |
|
Abstract
For Intelligent Assistants (IA), user activity is often used as a lag metric for user satisfaction or engagement. Conversely, predictive leading metrics for engagement can be helpful with decision making and evaluating changes in satisfaction caused by new features. In this paper, we propose User Return Time (URT), a fine grain metric for gauging user engagement. To compute URT, we model continuous inter-arrival times between users’ use of service via a log Gaussian Cox process (LGCP), a form of inhomogeneous Poisson process which captures the irregular variations in user usage rate and personal preferences typical of an IA. We show the effectiveness of the proposed approaches on predicting the return time of users on real-world data collected from an IA. Experimental results demonstrate that our model is able to predict user return times reasonably well and considerably better than strong baselines that make the prediction based on past utterance frequency.
|
| 8 |
Inductive Transfer Learning for Detection of Well-Formed Natural Language Search Queries |
Bakhtiyar Syed,Vijayasaradhi Indurthi,Manish Gupta,Manish Shrivastava,Vasudeva Varma |
|
Abstract
Users have been trained to type keyword queries on search engines. However, recently there has been a significant rise in the number of verbose queries. Often times such queries are not well-formed. The lack of well-formedness in the query might adversely impact the downstream pipeline which processes these queries. A well-formed natural language question as a search query aids heavily in reducing errors in downstream tasks and further helps in improved query understanding. In this paper, we employ an inductive transfer learning technique by fine-tuning a pretrained language model to identify whether a search query is a well-formed natural language question or not. We show that our model trained on a recently released benchmark dataset spanning 25,100 queries gives an accuracy of 75.03% thereby improving by .5 absolute percentage points over the state-of-the-art.
|
| 9 |
Towards Spatial Word Embeddings |
Paul Mousset,Yoann Pitarch,Lynda Tamine |
|
Abstract
Leveraging textual and spatial data provided in spatio-textual objects (eg., tweets), has become increasingly important in real-world applications, favoured by the increasing rate of their availability these last decades (eg., through smartphones). In this paper, we propose a spatial retrofitting method of word embeddings that could reveal the localised similarity of word pairs as well as the diversity of their localised meanings. Experiments based on the semantic location prediction task show that our method achieves significant improvement over strong baselines.
|
| 10 |
Asymmetry Sensitive Architecture for Neural Text Matching |
Thiziri Belkacem,Jose G. Moreno,Taoufiq Dkaki,Mohand Boughanem |
|
Abstract
Question-answer matching can be viewed as a puzzle where missing pieces of information are provided by the answer. To solve this puzzle, one must understand the question to find out a correct answer. Semantic-based matching models rely mainly in semantic relatedness the input text words. We show that beyond the semantic similarities, matching models must focus on the most important words to find the correct answer. We use attention-based models to take into account the word saliency and propose an asymmetric architecture that focuses on the most important words of the question or the possible answers. We extended several state-of-the-art models with an attention-based layer. Experimental results, carried out on two QA datasets, show that our asymmetric architecture improves the performances of well-known neural matching algorithms.
|
| 11 |
QGraph: A Quality Assessment Index for Graph Clustering |
Maria Halkidi,Iordanis Koutsopoulos |
|
Abstract
In this work, we aim to study the cluster validity problem for graph data. We present a new validity index that evaluates structural characteristics of graphs in order to select the clusters that best represent the communities in a graph. Since the work of defining what constitutes cluster in a graph is rather difficult, we exploit concepts of graph theory in order to evaluate the cohesiveness and separation of nodes. More specifically, we use the concept of ., and . to evaluate the connectivity of nodes . and . clusters. The effectiveness of our approach is experimentally evaluated using real-world data collections.
|
| 12 |
A Neural Approach to Entity Linking on Wikidata |
|
|
Abstract
|
| 13 |
Self-attentive Model for Headline Generation |
Daniil Gavrilov,Pavel Kalaidin,Valentin Malykh |
|
Abstract
Headline generation is a special type of text summarization task. While the amount of available training data for this task is almost unlimited, it still remains challenging, as learning to generate headlines for news articles implies that the model has strong reasoning about natural language. To overcome this issue, we applied recent Universal Transformer architecture paired with byte-pair encoding technique and achieved new state-of-the-art results on the New York Times Annotated corpus with ROUGE-L F1-score 24.84 and ROUGE-2 F1-score 13.48. We also present the new RIA corpus and reach ROUGE-L F1-score 36.81 and ROUGE-2 F1-score 22.15 on it.
|
| 14 |
Can Image Captioning Help Passage Retrieval in Multimodal Question Answering? |
Shurong Sheng,Katrien Laenen,Marie-Francine Moens |
|
Abstract
Passage retrieval for multimodal question answering, spanning natural language processing and computer vision, is a challenging task, particularly when the documentation to search from contains poor punctuation or obsolete word forms and with little labeled training data. Here, we introduce a novel approach to conducting passage retrieval for multimodal question answering of ancient artworks where the query image caption of the multimodal query is provided as additional evidence to state-of-the-art retrieval models in the cultural heritage domain trained on a small dataset. The query image caption is generated with an advanced image captioning model trained on an external dataset. Consequently, the retrieval model obtains transferred knowledge from the external dataset. Extensive experiments prove the efficiency of this approach on a benchmark dataset compared to state-of-the-art approaches.
|
| 15 |
A Simple Neural Approach to Spatial Role Labelling |
Nitin Ramrakhiyani,Girish Palshikar,Vasudeva Varma |
|
Abstract
Spatial Role Labelling involves identification of text segments which emit spatial semantics such as describing an object of interest, a reference point or the object’s relative position with the reference. Tasks in SemEval exercises of 2012 and 2013 propose problems and datasets for Spatial Role Labelling. In this paper, we propose a simple two-step neural network based approach to identify static spatial relations along with the three primary roles - Trajector, Landmark and Spatial Indicator. Our approach outperforms the task submission results and other state-of-the-art results on these datasets. We also include a discussion on the explainability of our model.
|
| 16 |
Neural Diverse Abstractive Sentence Compression Generation |
Mir Tafseer Nayeem,Tanvir Ahmed Fuad,Yllias Chali |
|
Abstract
In this work, we have contributed a novel abstractive sentence compression model which generates diverse compressed sentence with paraphrase using a neural . encoder decoder model. We impose several operations in order to generate diverse abstractive compressions at the sentence level which was not addressed in the past research works. Our model jointly improves the information coverage and abstractiveness of the generated sentences. We conduct our experiments on the human-generated abstractive sentence compression datasets and evaluate our system on several newly proposed Machine Translation (.) evaluation metrics. Our experiments demonstrate that the methods bring significant improvements over the state-of-the-art methods across different metrics.
|
| 17 |
Fully Contextualized Biomedical NER |
Ashim Gupta,Pawan Goyal,Sudeshna Sarkar,Mahanandeeshwar Gattu |
|
Abstract
Recently, neural network architectures have outperformed traditional methods in biomedical named entity recognition. Borrowed from innovations in general text NER, these models fail to address two important problems of polysemy and usage of acronyms across biomedical text. We hypothesize that using a fully-contextualized model that uses contextualized representations along with context dependent transition scores in CRF can alleviate this issue and help further boost the tagger’s performance. Our experiments with this architecture have shown to improve state-of-the-art F1 score on 3 widely used biomedical corpora for NER. We also perform analysis to understand the specific cases where our contextualized model is superior to a strong baseline.
|
| 18 |
DeepTagRec: A Content-cum-User Based Tag Recommendation Framework for Stack Overflow |
Suman Kalyan Maity,Abhishek Panigrahi,Sayan Ghosh,Arundhati Banerjee,Pawan Goyal,Animesh Mukherjee |
|
Abstract
In this paper, we develop a . based deep learning framework . to recommend appropriate question tags on Stack Overflow. The proposed system learns the content representation from question title and body. Subsequently, the learnt representation from heterogeneous relationship between user and tags is fused with the content representation for the final tag prediction. On a very large-scale dataset comprising half a million question posts, . beats all the baselines; in particular, it significantly outperforms the best performing baseline . achieving an overall gain of 60.8% and 36.8% in .@3 and .@10 respectively. . also achieves 63% and 33.14% maximum improvement in . and . respectively over ..
|
| 19 |
Document Performance Prediction for Automatic Text Classification |
Gustavo Penha,Raphael Campos,Sérgio Canuto,Marcos André Gonçalves,Rodrygo L. T. Santos |
|
Abstract
Query performance prediction (QPP) is a fundamental task in information retrieval, which concerns predicting the effectiveness of a ranking model for a given query in the absence of relevance information. Despite being an active research area, this task has not yet been explored in the context of automatic text classification. In this paper, we study the task of predicting the effectiveness of a classifier for a given document, which we refer to as document performance prediction (DPP). Our experiments on several text classification datasets for both categorization and sentiment analysis attest the effectiveness and complementarity of several DPP inspired by related QPP approaches. Finally, we also explore the usefulness of DPP for improving the classification itself, by using them as additional features in a classification ensemble.
|
| 20 |
Misleading Metadata Detection on YouTube |
Priyank Palod,Ayush Patwari,Sudhanshu Bahety,Saurabh Bagchi,Pawan Goyal |
|
Abstract
YouTube is the leading social media platform for sharing videos. As a result, it is plagued with misleading content that includes staged videos presented as real footages from an incident, videos with misrepresented context and videos where audio/video content is morphed. We tackle the problem of detecting such misleading videos as a supervised classification task. We develop UCNet - a deep network to detect fake videos and perform our experiments on two datasets - VAVD created by us and publicly available FVC [.]. We achieve a macro averaged F-score of 0.82 while training and testing on a 70:30 split of FVC, while the baseline model scores 0.36. We find that the proposed model generalizes well when trained on one dataset and tested on the other.
|
发表于 2025-3-21 19:53:32
