RESUMEN

Visual question generation aims to focus on some target objects in an image to generate questions with certain questioning purposes. Existing studies mainly utilize an answer to extract the target object corresponding to the questioning purpose for questioning. However, answers fail to accurately and completely map to every target object, such as the objects corresponding to the answer are ambiguous or the answers are the relationship between multiple objects. To address this problem, we propose a content-controlled question generation model, which generates questions based on a given target object set specified from an image. Considering that the target objects have different contributions during the generation process, we design a recurrent generative architecture to explicitly control attention to different objects and their corresponding image information at each generative stage. Extensive experiments on the VQA v2.0 dataset and the Visual7w dataset show that the proposed model outperforms the state-of-the-art models and can controllably generate questions with specified content.

RESUMEN

Distance learning frees the learning process from spatial constraints. Each mode of distance learning, including synchronous and asynchronous learning, has disadvantages. In synchronous learning, students have network bandwidth and noise concerns, but in asynchronous learning, they have fewer opportunities for engagement, such as asking questions. The difficulties associated with asynchronous learning make it difficult for teachers to determine whether students comprehend the course material. Motivated students will consistently participate in a course and prepare for classroom activities if teachers ask questions and communicate with them during class. As an aid to distance education, we want to automatically generate a sequence of questions based on asynchronous learning content. In this study, we will also generate multiple-choice questions for students to answer and teachers to easily correct. The asynchronous distance teaching-question generation (ADT-QG) model, which includes Sentences-BERT (SBERT) in the model architecture to generate questions from sentences with a higher degree of similarity, is proposed in this work. With the Wiki corpus generation option, it is anticipated that the Transfer Text-to-Text Transformer (T5) model will generate more fluent questions and be more aligned with the instructional topic. The results indicate that the questions created by the ADT-QG model suggested in this work have good fluency and clarity indicators, showing that the questions generated by the ADT-QG model are of a certain quality and relevant to the curriculum.

RESUMEN

In general, making evaluations requires a lot of time, especially in thinking about the questions and answers. Therefore, research on automatic question generation is carried out in the hope that it can be used as a tool to generate question and answer sentences, so as to save time in thinking about questions and answers. This research focuses on automatically generating short answer questions in the reading comprehension section using Natural Language Processing (NLP) and K-Nearest Neighborhood (KNN). The questions generated use article sources from news with reliable grammar. To maintain the quality of the questions produced, machine learning methods are also used, namely by conducting training on existing questions. The stages of this research in outline are simple sentence extraction, problem classification, generating question sentences, and finally comparing candidate questions with training data to determine eligibility. The results of the experiment carried out were for the Grammatical Correctness parameter to produce a percentage of 59.52%, for the Answer Existence parameter it yielded 95.24%, while for the Difficulty Index parameter it produced a percentage of 34.92%. So that the resulting average is 63.23%. So, this software deserves to be used as an alternative to automatically create reading comprehension questions.

RESUMEN

Auxiliary clinical diagnosis has been researched to solve unevenly and insufficiently distributed clinical resources. However, auxiliary diagnosis is still dominated by human physicians, and how to make intelligent systems more involved in the diagnosis process is gradually becoming a concern. An interactive automated clinical diagnosis with a question-answering system and a question generation system can capture a patient's conditions from multiple perspectives with less physician involvement by asking different questions to drive and guide the diagnosis. This clinical diagnosis process requires diverse information to evaluate a patient from different perspectives to obtain an accurate diagnosis. Recently proposed medical question generation systems have not considered diversity. Thus, we propose a diversity learning-based visual question generation model using a multi-latent space to generate informative question sets from medical images. The proposed method generates various questions by embedding visual and language information in different latent spaces, whose diversity is trained by our newly proposed loss. We have also added control over the categories of generated questions, making the generated questions directional. Furthermore, we use a new metric named similarity to accurately evaluate the proposed model's performance. The experimental results on the Slake and VQA-RAD datasets demonstrate that the proposed method can generate questions with diverse information. Our model works with an answering model for interactive automated clinical diagnosis and generates datasets to replace the process of annotation that incurs huge labor costs.

Asunto(s)

RESUMEN

We provide the first openly-available German QUestion-Answer Congruence Corpus (QUACC), designed for the task of sentence-based question generation with question-answer congruence. Based on this corpus, we establish suitable baselines for question generation, comparing systems of very different nature. Question generation is an interesting challenge in particular for current neural network architectures given that it combines aspects of language meaning and forms in complex ways. The systems have to generate question phrases appropriately linking to the meaning of the envisaged answer phrases, and they have to learn to generate well-formed questions using the source. We show that our QUACC corpus is well-suited to investigate the performance of various neural models and gain insights about the specific error sources.

RESUMEN

Question Generation (QG) aims to automate the task of composing questions for a passage with a set of chosen answers found within the passage. In recent years, the introduction of neural generation models has resulted in substantial improvements of automatically generated questions in terms of quality, especially compared to traditional approaches that employ manually crafted heuristics. However, current QG evaluation metrics solely rely on the comparison between the generated questions and references, ignoring the passages or answers. Meanwhile, these metrics are generally criticized because of their low agreement with human judgement. We therefore propose a new reference-free evaluation metric called QAScore, which is capable of providing a better mechanism for evaluating QG systems. QAScore evaluates a question by computing the cross entropy according to the probability that the language model can correctly generate the masked words in the answer to that question. Compared to existing metrics such as BLEU and BERTScore, QAScore can obtain a stronger correlation with human judgement according to our human evaluation experiment, meaning that applying QAScore in the QG task benefits to a higher level of evaluation accuracy.

RESUMEN

With the wide application of computers and digital technologies, online information searching is being integrated into students' learning process. Improving students' creative question generation through online information searching is an emerging research topic in the creativity and pedagogy field. Online information searching brings diversified information, but it also leads to cognitive load brought by a large amount of online information. Using online information searching to generate creative questions depends on students' cognitive properties. However, the existing literature ignores the joint influence of students' online information searching strategies and cognitive properties on their creative question generation. This study puts forward three hypotheses: first, the two strategies of students' online information searching ("keywords" and "Web page exploration") will increase their creative question generation; second, the impact of "keywords" is negatively moderated by students' need for cognitive closure (NFCC); third, the impact of "Web page exploration" is positively moderated by NFCC. The main reason is that high NFCC prevents students from obtaining diversified perspectives by using different keywords, but it helps to avoid distractions caused by a large amount of online information and promote the persistency of their reading information. Based on the data of quasi-experimental tasks completed by 90 students in Grade 7 and Grade 8, the results support the above hypothesis. The contributions of creative question generation theory and NFCC theory, as well as important issues of future study, are discussed.

RESUMEN

Background: Asking learners manually authored questions about their readings improves their text comprehension. Yet, not all reading materials comprise sufficiently many questions and many informal reading materials do not contain any. Therefore, automatic question generation has great potential in education as it may alleviate the lack of questions. However, currently, there is insufficient evidence on whether or not those automatically generated questions are beneficial for learners' understanding in reading comprehension scenarios. Objectives: We investigate the positive and negative effects of automatically generated short-answer questions on learning outcomes in a reading comprehension scenario. Methods: A learner-centric, in between-groups, quasi-experimental reading comprehension case study with 48 college students is conducted. We test two hypotheses concerning positive and negative effects on learning outcomes during the text comprehension of science texts and descriptively explore how the generated questions influenced learners. Results: The results show a positive effect of the generated questions on the participants learning outcomes. However, we cannot entirely exclude question-induced adverse side effects on learning of non-questioned information. Interestingly, questions identified as computer-generated by learners nevertheless seemed to benefit their understanding. Take Away: Automatic question generation positively impacts reading comprehension in the given scenario. In the reported case study, even questions recognized as computer-generated supported reading comprehension.

RESUMEN

Neural auto-regressive sequence-to-sequence models have been dominant in text generation tasks, especially the question generation task. However, neural generation models suffer from the global and local semantic semantic drift problems. Hence, we propose the hierarchical encoding-decoding mechanism that aims at encoding rich structure information of the input passages and reducing the variance in the decoding phase. In the encoder, we hierarchically encode the input passages according to its structure at four granularity-levels: [word, chunk, sentence, document]-level. Second, we progressively select the context vector from the document-level representations to the word-level representations at each decoding time step. At each time-step in the decoding phase, we progressively select the context vector from the document-level representations to word-level. We also propose the context switch mechanism that enables the decoder to use the context vector from the last step when generating the current word at each time-step.It provides a means of improving the stability of the text generation process during the decoding phase when generating a set of consecutive words. Additionally, we inject syntactic parsing knowledge to enrich the word representations. Experimental results show that our proposed model substantially improves the performance and outperforms previous baselines according to both automatic and human evaluation. Besides, we implement a deep and comprehensive analysis of generated questions based on their types.

RESUMEN

BACKGROUND: Generating questions by learners might be a potent learning technique but previous research yielded several shortcomings and underlying mechanisms are not well understood. METHODS: Students (N = 231) first read an expository text including bold keywords and then generated questions and answers referring to these keywords in three conditions: (1) open-book (i.e., text accessible), (2) closed-book (i.e., text inaccessible), and (3) cued closed-book (i.e., only keywords provided). RESULTS: In a test after one week, students in the open-book and in the cued closed-book conditions performed better than students in the restudying condition. The number of generated questions and answers was largest in the open-book condition, smaller in the cued closed-book condition and smallest in the closed-book condition and predicted final test performance. CONCLUSIONS: Generating questions and answers is an effective tool to boost retention in university learning when (at least part of) the learning material remains accessible.

Asunto(s)

RESUMEN

BACKGROUND: The ICAP framework based on cognitive science posits four modes of cognitive engagement: Interactive, Constructive, Active, and Passive. Focusing on the wide applicability of discussion as interactive engagement in medical education, we investigated the effect of discussion when it was preceded by self-study and further investigated the effect of generating questions before discussions. METHODS: This study was conducted in the second semester of 2018 and was participated in by 129 students majoring in health professions, including medicine, dentistry, veterinary medicine, and nursing. The students were assigned to four different trial groups and were asked to fill out a Subjective Mental Effort Questionnaire after completing each session. Their performance in posttest scores was analyzed using Bonferroni test, and mental effort was analyzed using mediation analysis. RESULTS: These results indicated that the self-study and question group had the highest performance and that the lecture and summary group had the lowest performance when comparing the total score. Using the analysis of mental effort, it was confirmed that the relationship between different study conditions and post-test performance was mediated by mental effort during test. CONCLUSIONS: Our findings support the ICAP framework and provide practical implications for medical education, representing the fact that students learn more when they are involved in active learning activities, such as self-study and question generation, prior to discussions.

Asunto(s)

RESUMEN

Objective. To investigate the degree to which student-generated questions or answering student-generated multiple-choice questions predicts course performance in medicinal chemistry. Methods. Students enrolled in Medicinal Chemistry III over a 3-year period were asked to create at least one question per exam period using PeerWise; within the software, they were also asked to answer and rate one peer question per class session. Students' total reputation scores and its components (question authoring, answering, and rating) and total answer scores (correctness of answers submitted indicating agreement with the author's chosen answer) were analyzed relative to final course grades. Results. Students at the non-satellite campus and those who generated more highly rated questions performed better overall in the course accounting for 12% of the variability in course grades. The most notable differences were between the top third and bottom third performing students within the course. The number of questions answered by students was not a significant predictor of course performance. Conclusion. Student generation of more highly rated questions (referred to as more thoughtful in nature by the software program) is predictive of course performance but it only explained a small variability in course grades. The correctness of answers submitted, however, did not relate to student performance.

Asunto(s)

RESUMEN

The present study investigates the best factor for controlling the item difficulty of multiple-choice English vocabulary questions generated by an automatic question generation system. Three factors are considered for controlling item difficulty: (1) reading passage difficulty, (2) semantic similarity between the correct answer and distractors, and (3) the distractor word difficulty level. An experiment was conducted by administering machine-generated items to three groups of English learners. The groups were determined based on their standardised English test scores. In total, 120 items, generated using combinations of the above three factors, were tested. The results reveal that the distractor word difficulty level had the greatest impact on item difficulty, but this tendency changed depending on the proficiency of the test takers. These results will be of use when implementing a fully automatic system for administrating tests.

RESUMEN

This paper describes details of the evaluation experiments for questions created by an automatic question generation system. Given a target word and one of its word senses, the system generates a multiple-choice English vocabulary question asking for the closest in meaning to the target word in the reading passage. Two kinds of evaluation were conducted considering two aspects: (1) measuring English learners' proficiency and (2) their similarity to the human-made questions. The first evaluation is based on the responses from English learners obtained through administering the machine-generated and human-made questions to them, and the second is based on the subjective judgement by English teachers. Both evaluations showed that the machine-generated questions were able to achieve a comparable level with the human-made questions in both measuring English proficiency and similarity.

RESUMEN

BACKGROUND: The increasing number of open-access ontologies and their key role in several applications such as decision-support systems highlight the importance of their validation. Human expertise is crucial for the validation of ontologies from a domain point-of-view. However, the growing number of ontologies and their fast evolution over time make manual validation challenging. METHODS: We propose a novel semi-automatic approach based on the generation of natural language (NL) questions to support the validation of ontologies and their evolution. The proposed approach includes the automatic generation, factorization and ordering of NL questions from medical ontologies. The final validation and correction is performed by submitting these questions to domain experts and automatically analyzing their feedback. We also propose a second approach for the validation of mappings impacted by ontology changes. The method exploits the context of the changes to propose correction alternatives presented as Multiple Choice Questions. RESULTS: This research provides a question optimization strategy to maximize the validation of ontology entities with a reduced number of questions. We evaluate our approach for the validation of three medical ontologies. We also evaluate the feasibility and efficiency of our mappings validation approach in the context of ontology evolution. These experiments are performed with different versions of SNOMED-CT and ICD9. CONCLUSIONS: The obtained experimental results suggest the feasibility and adequacy of our approach to support the validation of interconnected and evolving ontologies. Results also suggest that taking into account RDFS and OWL entailment helps reducing the number of questions and validation time. The application of our approach to validate mapping evolution also shows the difficulty of adapting mapping evolution over time and highlights the importance of semi-automatic validation.

Asunto(s)

RESUMEN

Discourse and argumentation are effective techniques for education not only in social domains but also in science domains. However, it is difficult for some teachers to stimulate an active discussion between students because several students might not be able to develop their arguments. This paper proposes to use WordNet as a semantic source in order to generate questions that are intended to stimulate students' brainstorming and to help them develop arguments in a discussion session. In a study including 141 questions generated by human experts and 44 questions generated by a computer system, the following research questions have been investigated: Are system-generated questions understandable? Are they relevant to given discussion topics? Would they be useful for supporting students in developing new arguments? Are understandable and relevant system-generated questions predicted to be useful for students in order to develop new arguments? The evaluation showed that system-generated questions could not be distinguished from human-generated questions in the context of two discussion topics while the difference between system-generated and human-generated questions was noticed in the context of one discussion topic. In addition, the evaluation study showed that system-generated questions that are relevant to a discussion topic correlate moderately with questions that are predicted as useful for students in developing new arguments in the context of two discussion topics and understandable system-generated questions are rated as useful in the context of one specific discussion topic.

RESUMEN

Recognizing the potential of online student question-generation to engage language learners in communicative activities and use the target language in a personally meaningful way for language and learning motivation development, an experimental study examining the English learning effects of this approach, in comparison to an online drill-and-practice strategy, was conducted. A quasi-experimental research design was adopted. Four sixth-grade classes (N = 106) participated in this study and were randomly assigned to different treatment groups. An online learning system supporting the various learning activities was adopted. The results of analysis of covariance (ANCOVAs) showed that students in the online student question-generation group performed significantly better in English assessments and exhibited higher learning motivation than those in the contrast group. The significance of this study and suggestions for instructional implementation and future works are also presented.

RESUMEN

Los mecanismos cognitivos de la generación de preguntas no son aún bien conocidos. Recientemente, se ha propuesto el modelo obstáculo-meta que asimila la comprensión a un proceso de resolución de un problema cognitivo, cuyos obstáculos hacia la meta pretendida dan lugar a las preguntas. Una predicción del modelo es la relación entre tipos de preguntas formuladas y tipos de representación mental que los sujetos manejan durante la comprensión de la información suministrada. Hasta ahora, esta predicción ha sido validada parcialmente y solo con información textual. En este trabajo se presenta un estudio empírico de validación más fiable de esta predicción, desarrollado en dos fases y que utiliza información no textual, como es el funcionamiento de dispositivos experimentales. Los resultados apoyan la predicción del modelo con suficiente potencia estadística.

The cognitive mechanisms underlying question generation are not yet well understood. Recently, the Obstacle-Goal model has been proposed. This model assimilates comprehension to a problem-solving cognitive process: the obstacles in the way to the intended goal originate the questions asked. A direct prediction from this model is the relationship between the distribution of types of questions asked, and the kinds of mental representations subjects elaborate to understand the provided information. Up to now, this prediction has been partially validated, and only textual information has been used to this purpose. In the present paper an empirical two-phase study is developed to validate the above prediction in a more reliable way, using non-textual information: the operation of experimental devices. Results support the prediction of the model with enough statistical power.

Asunto(s)