Textual Inference Research Articles

Text-Driven Video Prediction

Current video generation models usually convert signals indicating appearance and motion received from inputs (e.g., image and text) or latent spaces (e.g., noise vectors) into consecutive frames, fulfilling a stochastic generation process for the uncertainty introduced by latent code sampling. However, this generation pattern lacks deterministic constraints for both appearance and motion, leading to uncontrollable and undesirable outcomes. To this end, we propose a new task called Text-driven Video Prediction (TVP). Taking the first frame and text caption as inputs, this task aims to synthesize the following frames. Specifically, appearance and motion components are provided by the image and caption separately. The key to addressing the TVP task depends on fully exploring the underlying motion information in text descriptions, thus facilitating plausible video generation. In fact, this task is intrinsically a cause-and-effect problem, as the text content directly influences the motion changes of frames. To investigate the capability of text in causal inference for progressive motion information, our TVP framework contains a Text Inference Module (TIM), producing step-wise embeddings to regulate motion inference for subsequent frames. In particular, a refinement mechanism incorporating global motion semantics guarantees coherent generation. Extensive experiments are conducted on Something-Something V2 and Single Moving MNIST datasets. Experimental results demonstrate that our model achieves better results over other baselines, verifying the effectiveness of the proposed framework.

Comparative evaluation of image-based vs. text-based vs. multimodal AI approaches for automatic breast density assessment in mammograms

Background and objectivesIn the last decade, there has been a growing interest in applying artificial intelligence (AI) systems to breast cancer assessment, including breast density evaluation. However, few models have been developed to integrate textual mammographic reports and mammographic images. Our aims are (1) to generate a natural language processing (NLP)-based AI system, (2) to evaluate an external image-based software, and (3) to develop a multimodal system, using the late fusion approach, by integrating image and text inferences for the automatic classification of breast density according to the American College of Radiology (ACR) guidelines in mammograms and radiological reports. MethodsWe first compared different NLP models, three based on n-gram term frequency – inverse document frequency and two transformer-based architectures, using 1533 unstructured mammogram reports as a training set and 303 reports as a test set. Subsequently, we evaluated an external image-based software using 303 mammogram images. Finally, we assessed our multimodal system taking into account both text and mammogram images. ResultsOur best NLP model achieved 88 % accuracy, while the external software and the multimodal system achieved 75 % and 80 % accuracy, respectively, in classifying ACR breast densities. ConclusionAlthough our multimodal system outperforms the image-based tool, it currently does not improve the results offered by the NLP model for ACR breast density classification. Nevertheless, the promising results observed here open the possibility to more comprehensive studies regarding the utilization of multimodal tools in the assessment of breast density.

Recognizing Textual Inference in Mongolian Bar Exam Questions

This paper examines how to apply deep learning techniques to Mongolian bar exam questions. Several approaches that utilize eight different fine-tuned transformer models were demonstrated for recognizing textual inference in Mongolian bar exam questions. Among eight different models, the fine-tuned bert-base-multilingual-cased obtained the best accuracy of 0.7619. The fine-tuned bert-base-multilingual-cased was capable of recognizing “contradiction”, with a recall of 0.7857 and an F1 score of 0.7674; it recognized “entailment” with a precision of 0.7750, a recall of 0.7381, and an F1 score of 0.7561. Moreover, the fine-tuned bert-large-mongolian-uncased showed balanced performance in recognizing textual inference in Mongolian bar exam questions, thus achieving a precision of 0.7561, a recall of 0.7381, and an F1 score of 0.7470 for recognizing “contradiction”.

The Relationship of Text Inference Abilities and Reading Comprehension in Children with High-Functioning Autism Spectrum Disorder

Objectives: Reading comprehension in school-age children is an essential ability facilitates acquisition of information from general education. Yet, even with average or above cognitive and word recognition abilities, children with High Functioning Autism Spectrum Disorder (HF-ASD) demonstrate difficulties with reading comprehension. The purpose of the this study was to investigate that identifying the reading comprehension competence with a focus on text inference abilities (coherence processing, metalinguistic, theory of mind) in children with HF-ASD in the lower grades of elementary school.Methods: 15 HF-ASD and 15 age matched typically developing peers (TD) participated in this study. The participants performed reading comprehension tasks and text inferencing tasks (coherence processing, metalinguistic, theory of mind).Results: First, the reading comprehension accuracy of the HF-ASD group was significantly lower than the TD group. Second, the text inference accuracy of the HF-ASD group was significantly lower compared with the TD group. Third, in the TD group, there was a significant correlation between all text inference abilities (coherence processing, metalinguistic, theory of mind) and reading comprehension competence. In contrast, in the HF-ASD group, there was a positive correlation between coherence processing, metalinguistic and reading comprehension competence; however no significant relationship was found between theory of mind and reading comprehension competence. Finally, the coherence processing task predicted reading comprehension competence of children with HF-ASD by 52.2% and TD children by 42.1%.Conclusion: This study suggests that supporting connecting information within a text, local inferencing, and focusing on the coherence processing intervention is necessary for reading comprehension of children with HF-ASD in the lower grades of elementary school.

كلمة العدد (الإنجليزي)

كلمة العدد (الإنجليزي)

HUKUM ISLAM DAN DINAMIKA SOSIAL ‎

This research examines the renewal of contextual Islamic law discovery methods that function to answer the challenges of the times and social dynamics. As an initial step in the reconstruction of fiqh, this research seeks to explain a solution offered by fiqh methods, namely a unified approach to shari'ah and social inference. In simple terms, this method seeks to bridge and integrate textual (normative) and contextual (historical-empirical) approaches. The basic assumption is that this effort will be beneficial for alternative methods of discovering Islamic law in this multi-cultural and religious era. This research is a literature study (library research) using a juridical, historical, and philosophical approach to answer the problems above. This study concludes that there are three methods of ijtihad, namely bayani (linguistics), ta'lili (qiyasi: causation), and istislahi (teleological). These three are common methods used in discovering and forming fiqh civilizations from time to time. By bringing empirical reality into the analysis of legal findings, there will be little guarantee that Islamic law in Indonesia can appear more creative and alive in the midst of modern social regulatory processes. As for the effort to renew the method of discovering Islamic law with a contextual approach in the form of historical and textual inference analysis, this is an ideal intellectual achievement. there will be little guarantee that Islamic law in Indonesia can appear more creative and live in the midst of modern social regulatory processes. As for the effort to renew the method of discovering Islamic law with a contextual approach in the form of historical and textual inference analysis, this is an ideal intellectual achievement. there will be little guarantee that Islamic law in Indonesia can appear more creative and live in the midst of modern social regulatory processes. As for the effort to renew the method of discovering Islamic law with a contextual approach in the form of historical and textual inference analysis, this is an ideal intellectual achievement

Pragmatics of Phatic Utterances and Pleasantries in Nigerian Students Unionism

The apex body of students in any tertiary institution in Nigeria is student union which is led by students and for the students. Over the years, a unique pattern of greeting has evolved in Students Unionism. Student leaders tend to exhibit various phatic communions to develop and maintain social relationship. Extant studies on student union in the Nigerian context have largely addressed the phenomenon from the sociological and political perspectives with little attention paid to it from a linguistic perspective. This study, therefore, investigates the pragmatic functions of phatic utterances and pleasantries among student leaders in selected Nigerian tertiary institutions. Data was gathered from six tertiary institutions in the south-western part of Nigeria through observation and interview methods. Using Mey’s Pragmatic Act theory as the analytical tool, it was revealed that seven major functions are achieved with the phatic communion in Students Unionism. These include disclosure of political position, prayer invocation, placement by hierarchical status, assertion of loyalty and support, incitation of special recognition, issuance of warning and unveiling personality traits of student leaders. These are achieved through the deployment of pragmatic tools such as used are speech acts, physical acts and prosody in the activity part and in the textual part, reference (REF), inference (INF), metaphor (MPH) and Shared Social/Situation Knowledge (SSK) are deployed. These phatic communions are intentionally and uniquely used in the domain of Students Unionism to communicate a lot of messages in the achievement of their desired goal

From Text Signals to Simulations: A Review and Complement to Text as Data by Grimmer, Roberts & Stewart (PUP 2022)

Text as Data represents a major advance for teaching text analysis in the social sciences, digital humanities and data science by providing an integrated framework for how to conceptualize and deploy natural language processing techniques to enrich descriptive and causal analyses of social life in and from text. Here I review achievements of the book and highlight complementary paths not taken, including discussion of recent computational techniques like transformers, which have come to dominate automated language understanding and are just beginning to find their way into the careful research designs showcased in the book. These new methods not only highlight text as a signal from society, but textual models as simulations of society, which could fuel future advances in causal inference and experimentation. Text as Data's focus on textual discovery, measurement and inference points us toward this new frontier, cautioning us not to ignore, but build upon social scientific interpretation and theory.

BMCSA:Multi-feature spatial convolution semantic matching model based on BERT

This paper proposes a multi-feature spatial convolutional semantic matching model (BMCSA) based on BERT by enriching different feature spatial information of semantic features. BMCSA employs the BERT model to extract the semantic features of the text, then uses the two-dimensional convolutional network to extract different feature spatial information, and finally combines the Attention mechanism to capture the global feature spatial information. We use two different semantic matching data sets and a text inference data set to verify the effectiveness of the proposed model. Experimental results prove that BMCSA is better than the baseline model.

Enhancing pretrained language models with structured commonsense knowledge for textual inference

Transformer-based pretrained language models have shown promising performances in various textual inference tasks. However, additional relational knowledge between the semantic units within the input text may also play a critical role in the inference process. To equip the pretrained language models with the relational knowledge, previous methods employ a retrieval-based strategy, which obtain the relational features from prebuilt knowledge bases by a lookup operation. However, the inherent sparsity of part of the knowledge bases would prevent the direct retrieval of the relational features. To address this issue, we propose a MIX-strategy based Structural commonsense integration framework (Mix-Sci). In addition to the traditional retrieval strategy which only adapts to the knowledge bases with high coverage, Mix-Sci introduces an additional generative strategy to incorporate the sparse knowledge bases with the pretrained language models. In specific, in the training process, Mix-Sci learns to generate the structural information of the knowledge base, including the embedding of nodes and the connection relationship between the nodes. So that in the test process, the structural information can be generated to enhance the inference process. Experimental results on two textual inference tasks: machine reading comprehension and event prediction show that Mix-Sci can effectively utilize both the dense and the sparse knowledge bases, to consistently improve the performance of pretrained language models on textual inference tasks.

Cross-Modal Coherence for Text-to-Image Retrieval

Common image-text joint understanding techniques presume that images and the associated text can universally be characterized by a single implicit model. However, co-occurring images and text can be related in qualitatively different ways, and explicitly modeling it could improve the performance of current joint understanding models. In this paper, we train a Cross-Modal Coherence Model for text-to-image retrieval task. Our analysis shows that models trained with image–text coherence relations can retrieve images originally paired with target text more often than coherence-agnostic models. We also show via human evaluation that images retrieved by the proposed coherence-aware model are preferred over a coherence-agnostic baseline by a huge margin. Our findings provide insights into the ways that different modalities communicate and the role of coherence relations in capturing commonsense inferences in text and imagery.

Feature based Entailment Recognition for Malayalam Language Texts

Textual entailment is a relationship between two text fragments, namely, text/premise and hypothesis. It has applications in question answering systems, multi-document sum-marization, information retrieval systems, and social network analysis. In the era of the digital world, recognizing semantic variability is important in understanding inferences in texts. The texts are either in the form of sentences, posts, tweets, or user experiences. Hence understanding inferences from customer experiences helps companies in customer segmentation. The availability of digital information is ever-growing with textual data in almost all languages, including low resource languages. This work deals with various machine learning approaches applied to textual entailment recognition or natural language inference for Malayalam, a South Indian low resource language. A performance-based analysis using machine learning classification techniques such as Logistic Regression, Decision Tree, Support Vector Machine, Random Forest, AdaBoost, and Naive Bayes is done for the MaNLI (Malayalam Natural Language Inference) dataset. Different lexical and surface-level features are used for this binary and multiclass classification. With the increasing size of the dataset, there is a drop in the performance of feature-based classification. A comparison of feature-based models with deep learning approaches highlights this inference. The main focus here is the feature-based analysis with 14 different features and its comparison, essential to any NLP classification problem.

Textual Inference Comprehension in Mild Cognitive Impairment: The Influence of Semantic Processing and Verbal Episodic Memory.

Language complaints, especially in complex tasks, may occur in mild cognitive impairment (MCI). Various language measures have been studied as cognitive predictors of MCI conversion to Alzheimer's type dementia. Understanding textual inferences is considered a high-demanding task that recruits multiple cognitive functions and, therefore, could be sensitive to detect decline in the early stages of MCI. Thus, we aimed to compare the performance of subjects with MCI to healthy elderly in a textual inference comprehension task and to determine the best predictors of performance in this ability considering one verbal episodic memory and two semantic tasks. We studied 99 individuals divided into three groups: (1) 23 individuals with amnestic mild cognitive impairment (aMCI), (2) 42 individuals with non-amnestic mild cognitive impairment (naMCI), (3), and (4) 34 cognitively healthy individuals for the control group (CG). A reduced version of The Implicit Management Test was used to assess different types of inferential reasoning in text reading. MCI patients performed poorer than healthy elderly, and there were no differences between MCI subgroups (amnestic and non-amnestic). The best predictors for inference-making were verbal memory in the aMCI and semantic tasks in the naMCI group. The results confirmed that the failure to understand textual inferences can be present in MCI and showed that different cognitive skills like semantic knowledge and verbal episodic memory are necessary for inference-making.

Betalogger: Smartphone Sensor-based Side-channel Attack Detection and Text Inference Using Language Modeling and Dense MultiLayer Neural Network

With the recent advancement of smartphone technology in the past few years, smartphone usage has increased on a tremendous scale due to its portability and ability to perform many daily life tasks. As a result, smartphones have become one of the most valuable targets for hackers to perform cyberattacks, since the smartphone can contain individuals’ sensitive data. Smartphones are embedded with highly accurate sensors. This article proposes BetaLogger , an Android-based application that highlights the issue of leaking smartphone users’ privacy using smartphone hardware sensors (accelerometer, magnetometer, and gyroscope). BetaLogger efficiently infers the typed text (long or short) on a smartphone keyboard using Language Modeling and a Dense Multi-layer Neural Network (DMNN). BetaLogger is composed of two major phases: In the first phase, Text Inference Vector is given as input to the DMNN model to predict the target labels comprising the alphabet, and in the second phase, sequence generator module generate the output sequence in the shape of a continuous sentence. The outcomes demonstrate that BetaLogger generates highly accurate short and long sentences, and it effectively enhances the inference rate in comparison with conventional machine learning algorithms and state-of-the-art studies.

Statistical inference in abstracts of three influential clinical pharmacology journals analyzed using a text-mining algorithm.

To describe the trend in the prevalence of statistical inference in three influential clinical pharmacology journals METHODS: We applied a computer-based algorithm to abstracts of three clinical pharmacology journals published in 1976 to 2016 to identify statistical inference and its subtypes. Furthermore, we manually reviewed a random sample of 300 articles to access algorithm's performance in finding statistical inference in abstracts and as a screening tool for presence and absence of statistical inference in full text. The algorithm identified 59% (13,375/22,516 [mid p 95% CI, 59%-60%]) article abstracts with statistical inference. The percentage of abstracts with statistical inference was similar in 1976 and 2016, 48% (179/377 [mid p 95%CI, 42%-52%]) versus 49% (386/791 [mid p 95%CI, 45%-52%]). Statistical reporting pattern varied among journals. Among abstracts containing any statistical inference in the publications from 1976 to 2016 null-hypothesis significance testing was the most prevalent reported statistical inference. The algorithm had high sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for finding statistical inferences in abstract. While PPV for predicting the statistical inference in full text (including abstract, text, tables and figures) was high, NPV was low. Despite journal's editorials and statistical associations' guidelines, most authors focused on testing rather than estimation. In future, a better statistical reporting might be ensured by improving the statistical knowledge of authors and an addition of statistical guides to journals' instruction to authors to the extent that editors would like their statistical inference preferences to be incorporated into submitted manuscripts.

Single and Cross-Disorder Detection for Autism and Schizophrenia

Detection of mental disorders from textual input is an emerging field for applied machine and deep learning methods. Here, we explore the limits of automated detection of autism spectrum disorder (ASD) and schizophrenia (SCZ). We compared the performance of: (1) dedicated diagnostic tools that involve collecting textual data, (2) automated methods applied to the data gathered by these tools, and (3) psychiatrists. Our article tests the effectiveness of several baseline approaches, such as bag of words and dictionary-based vectors, followed by a machine learning model. We employed two more refined Sentic text representations using affective features and concept-level analysis on texts. Further, we applied selected state-of-the-art deep learning methods for text representation and inference, as well as experimented with transfer and zero-shot learning. Finally, we also explored few-shot methods dedicated to low data size scenarios, which is a typical problem for the clinical setting. The best breed of automated methods outperformed human raters (psychiatrists). Cross-dataset approaches turned out to be useful (only from SCZ to ASD) despite different data types. The few-shot learning methods revealed promising results on the SCZ dataset. However, more effort is needed to explore the approaches to efficiently training models, given the very limited amounts of labeled clinical data. Psychiatry is one of the few medical fields in which the diagnosis of most disorders is based on the subjective assessment of a psychiatrist. Therefore, the introduction of objective tools supporting diagnostics seems to be pivotal. This paper is a step in this direction.

Knowledge graph augmented advanced learning models for commonsense reasoning

Machine learning is the key solution to many AI issues, but learning models rely heavily on specific training data. While a Bayesian setup can be used to incorporate some learning patterns with previous knowledge, those patterns can not access any organized world knowledge on requirements. The primary objective is to enable human-capable machines in ordinary everyday circumstances to estimate and make presumptions. In this paper we propose to respond to such common sense issues through a textual inference system with external, organized common sense graphs for explanatory inferences. The framework is based on a schematic map as a pair of questions and answers, a linked subgraph from the semantine to the symbolic space of knowledge-based external information. It displays a schematic map with a new network graphic module for information knowledge and performance with graph representations. LSTMs and graphical networks with a hierarchical attention-based direction are the basis of our model. It is flexible and understandable from the intermediate attention scores, leading to confident results. We also achieved state-of-the-art reliability on CommonsenseQA, a broad database of common sense reasoning utilizing ConceptNet as the only external tool for BERT-based models.

Penentuan Awal Bulan Kamariah Untuk Ibadah

This paper presents data regarding the formula for determining the start of the month for worship at the time of the Prophet. This is important because the determination of the beginning of the month among Muslims has always been a polemic, even Muslims until now still do not have a definite and unique calendar to serve as a common guideline. The method usedin this research is the method introducedby Louay Safi "Unified Aproach to Shari'ah Inference"; namely textual normative inference, empirical-historical-contextualist inference, and integrated inference. The results were as follows: d natural inference stualis te k-normative-legis, fast orders clearly to be given information about ordinances that rukyatul hilal; and if it is cloudy or covered in fog, then it is ordered to calculate, estimate, or fulfill the month into 30 days and the rule of law in matters of mahdah worship such as fasting cannot be changed, sentences that are already text and sharih do not have the character of ijtihadi; and anything that creates controversy in society can be eliminated by the government or the authorities; whereas in the contextualist historical-empirical inference it can be said that worship which includes the obligation of rukyat has encouraged Muslims to study and deepen the science of astronomy ; and in unified inference, the computation of the reckoning at the time of the prophet with the fulfillment becomes thirty days because at that time the Madinans did not have zij or astronomical tables showing the positions of the moon, earth and sun. From this perspective, rukyat often called taabbudiy be part of the entity taaquliy for a silver lining can be found in the annals of the Muslims, and Reckoning ser ing said taaqquliy also part of ta'abbudiy entity because it is part and result of activity namely rukyat taabbudiy .

An Investigation about Entailment and Narrative by AI Techniques (Generative Models)

Many storytelling generation problems concern the difficulty to model the sequence of sentences. Language models are generally able to assign high scores to well-formed text, especially in the cases of short texts, failing when they try to simulate human textual inference. Although in some cases output text automatically generated sounds as bland, incoherent, repetitive and unrelated to the context, in other cases the process reveals capability to surprise the reader, avoiding to be boring/predictable, even if the generated text satisfies entailment task requirements. The lyric tradition often does not proceed towards a real logical inference, but takes into account alternatives like the unexpectedness, useful for predicting when a narrative story will be perceived as interesting. To achieve a best comprehension of narrative variety, we propose a novel measure based on two components: inference and unexpectedness, whose different weights can modify the opportunity for readers to have different experiences about the functionality of a generated story. We propose a supervised validation treatment, in order to compare the authorial original text, learned by the model, with the generated one.

Deep Learning and NLP based Side Channel Attack for Text Inference in Smartphones

Over the past years, smartphones have witnessed an alarming rise in embedded sensors which enhance their support for applications. However, they can be regarded as loopholes as seemingly innocuous information can be obtained without any user permissions in Android thus invading the user’s privacy. Our work establishes a side channel attack by illegitimately inferring the information being typed by the user on a smartphone using the readings from ‘zero-permission’ sensors like accelerometer and gyroscope. This serves as a proof of concept to prevent such attacks on mobile devices in the future. While previous research has been conducted in this space, our narrative involves a predictive model using Recurrent Neural Networks that can predict the letters being typed in the keyboard solely based on the motion sensor readings, thus inferring the text. Our research was able to identify 37.5% of the unseen words typed by the user using a very small volume of training data. Our tap detection method has shown 92% accuracy which plays a critical role in the text inference. This research lays the foundation to further progress in this area, thus helping to strengthen the mobile security.