Text Answers Research Articles

One-week multidisciplinary post-graduate palliative care training: an outcome-based program evaluation

BackgroundA multi-professional, post-graduate, one-week palliative care training program was piloted in November 2019 at the University of Ivano-Frankivsk, Ukraine. A formal evaluation of this program was performed.MethodsThis is a comparative, retrospective outcome-based evaluation of an educational intervention. Participants completed evaluation forms at the end of the course (post-intervention = T1), covering demographics, comparative retrospective self-assessment (40 items, 6-point Likert scale), organizational aspects, and general feedback (free text). At T1, the responses represent actual self-assessment, pre-interventional (T0) scores were generated by retrospective self-assessment. The Retrospective Performance Gain (RPG) was calculated on group level for the comparative self-assessment, demographic and organizational variables were analyzed by descriptive statistics, and free text answers were processed by qualitative methodology (content analysis).ResultsFifty-three of 56 attendants from all professions relevant to palliative care completed the evaluation forms (response 94,6%), with mean age 39y (22–64) and mean working experience 13,6y (1–44). Overall ratings of the program were very positive. Comparative retrospective self-assessment demonstrated a marked RPG from T0 to T1 on all items. Free text comments emphasized the need for regular nation-wide educational programs and for further education in bereavement care; inter-professional practice; communication; palliative care philosophy; professional self-care; specific nursing skills; dementia care; and advocacy, while the general contribution of the program to palliative care development in Ukraine was acknowledged.ConclusionsSystematic evaluation of a post-graduate international training program in palliative care may provide a mutual learning experience and map country-specific barriers and facilitators that have to be addressed when setting up palliative care services.

Women's ideal and real expectations of postnatal care during their first pregnancy: An online survey in England

BackgroundThere are many studies of women's experiences of care during the postnatal period, however little is known about women's expectations of postnatal care. ObjectiveThis study explores first-time pregnant women's expectations, both ideal and real life, of postnatal care in England. Designa descriptive, cross-sectional online survey design was used. The questionnaire took approximately 10 minutes to complete and was developed specifically for this survey. It included an informed consent section, socio-demographic questions and closed tick-box questions on where they had received information on postnatal care, and real and ideal expectations of postnatal care in hospital/birth centre and at home. SettingThe survey was hosted on the National Perinatal Epidemiology Unit website and advertised through a number of third sector and commercial organisations in 2017. ParticipantsWomen who were pregnant, had not given birth before, were aged 16 years and over, and living in England were eligible to participate. AnalysisSurvey data were analysed using descriptive statistics and, where appropriate, chi square test using SPSS Version 23. Data from open ended questions were analysed by two researchers separately then codes and themes were discussed until consensus was reached. Results283 women responded to the survey of whom 200 were eligible and included in the analysis. Most had received information on postnatal care from multiple sources, with pregnancy classes and midwives being most common. Most expected to stay one day or less in hospital or birth centre after normal delivery. Real life expectations were lower than ideal expectations, and hospital/birth centre real life expectations were higher than home real life expectations for physical health advice/checks and information/help with feeding. Categories developed from the open text answers were 'Respect, compassion and individualised care at a vulnerable time', 'The ward environment', 'Feeling ready for hospital discharge' and 'Help to find support in the community'. Key conclusionsWomen in this survey had high ideal world expectations of their postnatal care but in real life expected more focus on checking on their health and that of their baby and on giving information about the new challenges of how to breastfeed and look after a baby. While women valued checks of their health and that of their baby, ideally they wanted easy access to reassurance that they were feeding and looking after their baby well, that they were ‘doing it right’, and that what was happening to them was normal. Implications for practiceAs well as the necessary checks in the immediate postpartum period, consideration also needs to be given to the best way to meet the informational and support needs of women to optimise their wellbeing and transition to parenthood. A number of resources are used by women that could be enhanced to inform expectations of postnatal care and to provide valuable information to support their postnatal care.

Developmental problems concerning children's oral functions, based on a questionnaire administered to dentists and guardians

This study aimed to clarify the importance of dental support for eating problems of infants, young children and disabled children, based on a questionnaire survey administered to dentists and guardians. The respondents consisted of 1001 dentists, pediatric dentistry administrators randomly selected by each prefectural or municipal dental association, and 1031 guardians of preschoolers who attended kindergartens or children's day-care centers. Answers were received from 712 dentists and 844 guardians. We compared answers to question No. 16 for dentists and questions No. 23 & 26 for guardians. Furthermore, we compared free comments to question No. 16 for dentists and answers to No. 27 for guardians. Question No. 27 for dentists and question No. 21 for guardians similarly asked about children's eating problems, and we thus next compared the tendencies noted in answers to these questions. In addition, a componential analysis was conducted for free text answers to question No. 31 for dentists and question No. 34 for guardians to qualitatively examine “free comments to children's eating problems”. Free comments were statistically analyzed using a text-mining approach in IBM SPSS Text Analytics for Surveys. The results revealed that guardians had more concerns about the following issues (relative to dentists): “vegetables” (23-fold) and “picky eating/fad eating/biased eating” (eight-fold). Issues of more concern to dentists (relative to guardians) were “mastication/chewing” (two-fold) and “dental caries/decay” (1.3-fold). This study suggested that guardian's anxieties regarding children's eating problems have variable background factors.

360-degree Delphi: addressing sociotechnical challenges of healthcare IT

BackgroundIT systems in the healthcare field can have a marked sociotechnical impact: they modify communication habits, alter clinical processes and may have serious ethical implications. The introduction of such systems involves very different groups of stakeholders because of the inherent multi-professionalism in medicine and the role of patients and their relatives that are often underrepresented. Each group contributes distinct perspectives and particular needs, which create specific requirements for IT systems and may strongly influence their acceptance and success. In the past, needs analysis, challenges and requirements for medical IT systems have often been addressed using consensus techniques such as the Delphi technique. Facing the heterogeneous spectrum of stakeholders there is a need to develop these techniques further to control the (strong) influence of the composition of the expert panel on the outcome and to deal systematically with potentially incompatible needs of stakeholder groups.This approach uses the strong advantages a Delphi study has, identifies the disadvantages of traditional Delphi techniques and aims to introduce and evaluate a modified approach called 360-Degree Delphi. Key aspects of 360-Degree Delphi are tested by applying the approach to the needs and requirements analysis of a system for managing patients’ advance directives and living wills.Methods360-Degree Delphi (short 360°D), as a modified Delphi process, is specified as a structured workflow with the optional use of stakeholder groups. The approach redefines the composition of the expert panel by setting up groups of different stakeholders. Consensus is created within individual stakeholder groups, but is also communicated between groups, while the iterative structure of the Delphi process remains unchanged. We hypothesize that (1) 360-Degree Delphi yields complementary statements from different stakeholders, which would be lost in classical Delphi; while (2) the variation of statements within individual stakeholder groups is lower than within the total collective. A user study is performed that addresses five stakeholder groups (patients, relatives, medical doctors, nurses and software developers) on the topic of living will communication in an emergency context. Qualitative open questions are used in a Delphi round 0. Answer texts are coded by independent raters who carry out systematic bottom-up qualitative text analysis. Inter-rater reliability is calculated and the resulting codes are used to test the hypotheses. Qualitative results are transferred into quantitative questions and then surveyed in round 1. The study took place in Germany.ResultsAbout 25% of the invited experts (stakeholders) agreed to take part in the Delphi round 0 (three patients, two relatives, three medical doctors, two qualified nurses and three developers), forming a structured panel of the five stakeholder groups. Two raters created a bottom-up coding, and 238 thematic codes were identified by the qualitative text analysis. The inter-rater reliability showed that 44.95% of the codes were semantically similar and coded for the same parts of the raw textual replies. Based on a consented coding list, a quantitative online-questionnaire was developed and send to different stakeholder groups.With respect to the hypotheses, Delphi round 0 had the following results: (1) doctors had a completely different focus from all the other stakeholder groups on possible channels of communications with the patient; (2) the dispersion of codes within individual stakeholder groups and within the total collective – visualized by box plots – was approximately 28% higher in the total collective than in the sub-collectives, but without a marked effect size. With respect to the hypotheses, Delphi round 1 had the following results: different stakeholder groups had highly diverging opinions with respect to central questions on IT-development. For example, when asked to rate the importance of access control against high availability of data (likert scale, 1 meaning restrictive data access, 6 easy access to all data), patients (mean 4.862, Stdev +/− 1.866) and caregivers (mean 5.667, Stdev: +/− 0.816) highly favored data availability, while relatives would restrict data access (mean 2.778, stdev +/− 1.093). In comparison, the total group would not be representative of either of these individual stakeholder needs (mean 4.344, stdev +/− 1.870).Conclusion360-Degree Delphi is feasible and allows different stakeholder groups within an expert panel to reach agreement individually. Thus, it generates a more detailed consensus which pays more tribute to individual stakeholders needs. This has the potential to improve the time to consensus as well as to produce a more representative and precise needs and requirements analysis. However, the method may create new challenges for the IT development process, which will have to deal with complementary or even contradictory statements from different stakeholder groups.

Understanding residents’ perceptions of nature and local economic activities using an open-ended question before protected area designation in Amami Islands, Japan

For successful protected area (PA) management, it is essential to understand residents’ perceptions during the early phases of the designation process. However, most studies on residents’ perceptions have been conducted after PA designation due to the lack of researcher–policymaker cooperation. In this study, we reveal residents’ perceptions before the PA designation of Japan’s Amami Islands as a national park and a Natural World Heritage Site. We conducted a questionnaire survey using an open-ended question to collect textual answers on residents’ perceptions of nature and the local economic activities. We then categorized these answers into six topics by applying topic models and interpreted the topics qualitatively, indicating their content. We also examined the relativity of the topics and the islands using correspondence analysis.The residents were more interested in the landscapes relevant to their livelihoods and expected them to be managed. This result implies discrepancy between residents’ perceptions and the PA draft management plan because the draft plan mainly focuses on the conservation of biodiversity in subtropical rainforests, whereas residents were unfamiliar with this. With regard to the local economic activities, residents expected enhancement of agriculture and traditional craft industries and nature-based tourism. Furthermore, residents’ perceptions were probably influenced by the context of the islands on which they lived. We suggest adoption of suitable PA management and communication strategies for each island in view of residents’ perceptions. Our approach has enabled us to understand residents’ perceptions that have been disregarded through the PA designation process.

Joint Learning of Answer Selection and Answer Summary Generation in Community Question Answering

Community question answering (CQA) gains increasing popularity in both academy and industry recently. However, the redundancy and lengthiness issues of crowdsourced answers limit the performance of answer selection and lead to reading difficulties and misunderstandings for community users. To solve these problems, we tackle the tasks of answer selection and answer summary generation in CQA with a novel joint learning model. Specifically, we design a question-driven pointer-generator network, which exploits the correlation information between question-answer pairs to aid in attending the essential information when generating answer summaries. Meanwhile, we leverage the answer summaries to alleviate noise in original lengthy answers when ranking the relevancy degrees of question-answer pairs. In addition, we construct a new large-scale CQA corpus, WikiHowQA, which contains long answers for answer selection as well as reference summaries for answer summarization. The experimental results show that the joint learning method can effectively address the answer redundancy issue in CQA and achieves state-of-the-art results on both answer selection and text summarization tasks. Furthermore, the proposed model is shown to be of great transferring ability and applicability for resource-poor CQA tasks, which lack of reference answer summaries.

Transfer learning for multimodal dialog

Audio-Visual Scene-Aware Dialog (AVSD) is best understood as an extension of Visual Question Answering, the task of generating a textual answer in response to a textual question on multi-media content. In AVSD, the answer-relevant “context” is expanded to include past dialog turns, which we view as a specialized form of extra textual knowledge (in addition to the standard video features). We have developed a framework that uses hierarchical attention to fuse contributions from different modalities, and had shown how it can be used to generate textual summaries from multi-modal sources, specifically videos with accompanying commentary. In this paper, we transfer the algorithmic approach, models, and data from this background corpus of 2000 h of how-to videos to the AVSD task, and report our findings. Our approach uses dialog context, but makes no assumption about the ordering of the history. Our system achieves the best performance in both automatic and human evaluations in the 7th Dialog State Tracking Challenge (AVSD).

Answer Selection Using Multi-Layer Semantic Representation Learning

Answer selection is the most important module of question answering system. Mining the semantic relevance between questions and answers is the key point of answer selection. However, the answer texts often contain some colloquial expressions and redundant information in some fields. In these cases, some traditional methods of mining text features are complex and the performance improvement is limited. In this paper we investigate using deep learning to enhance the computability of semantics and eliminate the semantic gap for the answer selection module of non-factual question answering system. By integrating different networks to mine the association semantic between questions and answers, a novel multi-layer answer selection model is proposed. The shallow semantic relevance between questions and answers is mined by adding the attention mechanism to CNN, and then the features are input into Bi-LSTM model for further mining to obtain more efficient semantic representation. The experimental results show that compared with other models, our method can achieve the better accuracy in the task of answer selection.

USA Department of Defense audit of surgical antibiotic prophylaxis prescribing patterns in inguinal hernia repair.

Antibiotic prophylaxis in inguinal hernia repair (IHR) is contentious in literature and practice. In low-risk patients, for whom evidence suggests antibiotic prophylaxis is unnecessary, many surgeons still advocate for its routine use. This study surveys prescription patterns of Department of Defense (DoD) general surgeons. An anonymous survey was sent electronically to approximately 350 DoD general surgeons. The survey asked multiple-choice and free text answers about prescribing patterns and knowledge of current evidence for low-risk patients undergoing elective open inguinal hernia repair without mesh (OIHRWOM), open inguinal hernia repair with mesh (OIHRWM), or laparoscopic inguinal hernia repair (LIHR). 110 DoD general surgeons consented to participate. 58.6, 95 and 84.2% of surgeons always administer antibiotic prophylaxis in OIHRWOM, OIHRWM, and LIHR, respectively. 37.9, 70.9, and 63.2% of surgeons believe that it reduces rates of surgical site infection in OIHRWOM, OIHRWM, and LIHR, respectively. The most common reasons for empirically prescribing antibiotic prophylaxis include "I think the evidence supports it" (27 of 72 responses), "I would rather be conservative and safe" (15 of 72 responses), and "I am following my hospital/department guidelines" (9 of 72 responses). 11.8, 40.8, and 32.9% of surgeons believe current evidence supports antibiotic prophylaxis use in OIHRWM, OIHRWOM, and LIHR, respectively. 50, 18.4, and 22.4% of surgeons believe current evidence refutes antibiotic prophylaxis use in OIHRWM, OIHRWOM, and LIHR, respectively. The survey results indicate that the majority of practicing DoD general surgeons still empirically prescribe surgical antibiotic prophylaxis in IHR despite more conflicting opinions that it has no meaningful effect or that current evidence does not supports its use.

Investigating brand strategy gap in Hong Kong: first impressions and unique attractions

ABSTRACT A brand strategy gap refers to a deviation of the projected destination image of the destination marketing organizations (DMOs) from the perceived image of residents. This gap can be represented by the perceptions of the city’s most salient features and unique attractions by government officials and local residents. Given the difference in sample sizes, the gap is examined with the aid of CATPAC, a neutral network content analysis software which allows the quantification of the frequency and clustering of textual answers of the respondent groups, and the relationships across such answers. In Hong Kong, an empirical study confirms that the two groups share similarities and discrepancies regarding both the most salient destination images and unique attractions. They share characteristics of an urban destination as the first impression, and features (Victoria Harbour, cultural attributes, skyscrapers and beaches) that are unique attractions though respondents also create different mental interrelationships in their perceptions of the destination image.

Abstract P2-12-10: Joven & Fuerte (J&F): Program for young women with breast cancer in Mexico strongly aids in supporting young patients’ needs and easing their illness process

Abstract Background: Mexican young women with breast cancer (YWBC) are a distinct population with particular characteristics and special needs that are not entirely known and are therefore not systematically addressed. J&F is the first program of its kind in Latin America and intends to meet the needs of YWBC in terms of information delivery and support on issues such as fertility, genetics and psychology, as well as to characterize this group by establishing a prospective cohort. The aim of this study was to evaluate patients’ outlook and satisfaction with the services provided by J&F to determine its strengths and areas of opportunity in order to implement strategies to better fulfill patients’ needs. Methods: This cross-sectional study analyzed patients’ satisfaction with J&F at Instituto Nacional de Cancerologia at Mexico City and TecSalud at Monterrey. For this matter, an online survey, kindly provided by the PYNK program, was adapted and piloted to fit the assessment of the J&F provided services. An invitation to anonymously complete it was offered via a messaging application to patients who had been enrolled in J&F for ≥6 months. The questionnaire included multiple choice, Likert scale and open-ended questions exploring patients’ demographics and evaluation of the program’s information delivery and support services, as well as their perspectives of the program’s research component. Likert scale answers were dichotomized to ease analysis, establishing a cut-off point according to their distribution; and free text answers were codified after selecting the most frequently addressed topics. Descriptive statistics, t-student and ANOVA were used for analysis. Results: Of 484 eligible patients, 134 (28%) agreed to answer the survey. Median age at diagnosis was 35 years. A considerable proportion had higher education (78%), were unpartnered (37%) and childless (26%). The timing, clarity and usefulness of the information provided by J&F were each described as good/very good by >80% for the written format and ≥98% for the verbal one. More than 82% of patients reported being very satisfied (≥9/10) with the psychological support, genetic assessment and support groups; and 74%, with the educational workshops and the Whatsapp patients’ group. The program overall was useful/very useful according to 97% of patients and aided 82% to better cope with their illness. Patients found that the most useful services were psychological support, information resources and support groups, for 36%, 19% and 11%, respectively. The number of support services used by patients was significantly associated with their perception of the program’s usefulness (p=0.008) and its aid in easing their illness process (p=0.009). Regarding fertility issues, 81% of patients recalled being informed about the potential effects of cancer treatment on fertility and 45% about preservation strategies. Notably, 33% of patients wished to have more children at diagnosis and 29% wanted to preserve fertility, with J&F supporting 20% of them to undergo a preservation procedure. Patients reported that J&F could improve by offering more educational workshops (31%) and by giving more dissemination to its support services (21%). Finally, 58% of patients were unaware of the J&F ongoing research component. Conclusions: Patient satisfaction with J&F is very high, reflecting that the program is achieving its main goals by meeting Mexican YWBC’s needs by providing clear and useful means of information and support services in a low-resource setting. Also, it is relevant to promote J&F services and highlight its research component to enhance the program’s supportive reach. It is imperative to keep efforts to guarantee the program’s continuity and advocate for the extension of its benefits to other oncologic centers. Citation Format: Cynthia Villarreal-Garza, Alejandra Platas, Melina Miaja, Marisol Garcia-Garcia, Fernanda Mesa-Chavez, Alan Fonseca, Claudia Pineda, Sylvia de la Rosa-Pacheco, Enrique Bargallo-Rocha, David Rodriguez-Gomez, Alejandro Mohar. Joven & Fuerte (J&F): Program for young women with breast cancer in Mexico strongly aids in supporting young patients’ needs and easing their illness process [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P2-12-10.

Medical training in palliative care at the Hannover Medical School: development of the interdisciplinary and cross-sectoral education in the cross-sectional subject Q13 (WEISE-Q13)

The implementation of the cross-sectional subject of palliative care (Q13) at medical faculties in Germany is heterogeneous. In faculties without achair for palliative medicine, other solutions are needed. This study describes the development and implementation process of the new educational concept in Q13 at the Hannover Medical School and its evaluation by medical students and lecturers. The new educational concept was developed in 2017/18 following the six-step approach of curriculum development by Kern, and was implemented and tested in the academic year 2018/19. Evaluation was assessed in two steps, both in qualitative and quantitative terms (mixed methods): an anonymous written evaluation by students with the option of free text answers and feedback discussions with lecturers. Students and lecturers welcomed the new educational concept in equal measures. Students thought that education in Q13 stimulates reflection on the individual attitude towards the issues of death and dying and prepares them for encounters with terminally ill and dying people. The case-based learning units in particular led to aconsolidation of theoretical knowledge (i.e. pain management) with adequate patient orientation. Teachers positively highlighted the organization and the educational material. Content structuring was considered to be particularly helpful. With the new teaching concept, medical students feel well prepared for the treatment of terminally ill and dying patients. The combination of different learning formats with an interdisciplinary and cross-sectoral focus promotes education with apractical orientation.

Identifying Experts in Community Question Answering Website Based on Graph Convolutional Neural Network

High-quality answers, usually given by experts, play an important role in community question answering (CQA) websites. Therefore, experts in these websites are defined as those who provide high-quality answers. We propose two semi-supervised learning models based on the graph convolution neural network (GCN) to identify them. Both models comprehensively extract features from the social behavior network, user profiles, and question and answer text. Specifically, we construct a social behavior network according to the co-answering relationship among answerers, which means every two answerers are connected if they answer the same questions. The difference between these two models is the methods of extracting text features. One model named GCN-Doc uses Doc2vec to get text vectors before training. The other model named GCN-Lstm with a long short term memory (LSTM) network extracts text features while training. Experiments using real-world data from Zhihu.com, one of the largest Chinese CQA websites, show that both GCN-Doc and GCN-Lstm can identify experts effectively comparing with baselines of PageRank and other GCN based neural network models. Besides, GCN-Lstm performs better than GCN-Doc.

ACLSTM：A Novel Method for CQA Answer Quality Prediction Based on Question-answer Joint Learning

Given the limitations of the community question answering (CQA) answer quality prediction method in measuring the semantic information of the answer text, this paper proposes an answer quality prediction model based on the question-answer joint learning (ACLSTM). The attention mechanism is used to obtain the dependency relationship between the Question-and-Answer (Q&A) pairs. Convolutional Neural Network (CNN) and Long Short-term Memory Network (LSTM) are used to extract semantic features of Q&A pairs and calculate their matching degree. Besides, answer semantic representation is combined with other effective extended features as the input representation of the fully connected layer. Compared with other quality prediction models, the ACLSTM model can effectively improve the prediction effect of answer quality. In particular, the mediumquality answer prediction, and its prediction effect is improved after adding effective extended features. Experiments prove that after the ACLSTM model learning, the Q&A pairs can better measure the semantic match between each other, fully reflecting the model’s superior performance in the semantic information processing of the answer text.

Textual and pictorial representations in strategic thinking about linguistic meaning

The aim of this study was to investigate L2 speakers’ ability to think strategically about linguistic meaning by asking them to make sense of particle verb (PV) constructions, a particularly demanding aspect of the English language for L2 speakers. Our focus was on meaning construal strategies in textual and pictorial representations of 22 figurative PVs with the particle down. The participants were asked to express themselves verbally and visually, and we were interested in the nature of their answers as well as their relationship. More specifically, we wished to determine the salience of particular elements in the participants’ strategic meaning construal, the type of relationship between textual and pictorial representations and, finally, potential dominance of one mode over another, which was examined in terms of the well-established concepts of “relay” and “anchorage”. The results showed that participants generally related the meaning of the PV construction to its components in their textual answers, whereas in pictorial answers their main tendency was to attend to the figurative meaning of the PV. Furthermore, their textual and pictorial answers most frequently depended on each other, which allowed us to determine that the text-picture relationship was predominantly that of relay, i.e. that text was perceived as more significant in the text-picture relationship.

ORDENAÇÃO E SEQUENCIAÇÃO DE PERGUNTAS DE LEITURA: PROPOSTA DE TRABALHO PARA CRÔNICA E PINTURA NO LIVRO DIDÁTICO

RESUMO Analisam-se os tipos de perguntas e sua sequenciação, em uma coleção de livros didáticos do 6.º ao 9.º ano, na seção “Cruzando Linguagens”, que apresenta perguntas de leitura para gêneros diferentes, mas com temáticas próximas. A partir disso, propõe-se nova sequenciação de perguntas para uma das atividades do livro do 6º ano, destinada aos gêneros crônica e pintura. O trabalho centraliza-se na perspectiva sobre leitura interacionista, nos estudos da Linguística Aplicada e da prática de ordenação e de sequenciação de leitura, discutida por Solé (1998) e ampliada por Menegassi (2008, 2010a, 2010b, 2010c, 2010d, 2016), Fuza e Menegassi (2018, 2017) e Fuza (2017). Dentre as questões da seção, geralmente, apenas uma delas possibilita o diálogo temático entre textos, sendo assim, sugeriu-se: análise das perguntas do livro, voltadas à crônica e pintura; nova sequenciação das perguntas; verificação das lacunas temáticas. Na sequência, desenvolveu-se a proposta teórico-metodológica de ordenação e sequenciação de perguntas, com foco no diálogo entre os textos, que possibilita ao leitor: estabelecer o diálogo entre textos com aproximação temática; elaborar respostas textuais, inferenciais e interpretativas, envolvendo as ideias principais dos textos; produzir sentidos quanto às temáticas apresentadas, alcançando as compreensões textuais e inferenciais e a interpretação, possibilitando tratar da exauribilidade temática relativa.

A Machine Learning Approach for Automated Evaluation of Short Answers Using Text Similarity Based on WordNet Graphs

Answer sheet evaluation is a time-consuming task that requires lot of efforts by the teachers and hence there is a strong need of automation for the same. This paper proposes a machine learning based approach that relies on WordNet graphs for finding out the text similarity between the answer provided by the student and the ideal answer provided by the teacher to facilitate the automation of answer sheet evaluation. This work is the first attempt in the field of short answer-based evaluation using WordNet graphs. Here, a novel marking algorithm is provided which can incorporate semantic relations of the answer text into consideration. The results when tested on 400 answer sheets yield promising results as compared with the state-of-art.

KLOSURE: Closing in on open\u2013ended patient questionnaires with text mining

BackgroundKnee injury and Osteoarthritis Outcome Score (KOOS) is an instrument used to quantify patients’ perceptions about their knee condition and associated problems. It is administered as a 42-item closed-ended questionnaire in which patients are asked to self-assess five outcomes: pain, other symptoms, activities of daily living, sport and recreation activities, and quality of life. We developed KLOG as a 10-item open-ended version of the KOOS questionnaire in an attempt to obtain deeper insight into patients’ opinions including their unmet needs. However, the open–ended nature of the questionnaire incurs analytical overhead associated with the interpretation of responses. The goal of this study was to automate such analysis. We implemented KLOSURE as a system for mining free–text responses to the KLOG questionnaire. It consists of two subsystems, one concerned with feature extraction and the other one concerned with classification of feature vectors. Feature extraction is performed by a set of four modules whose main functionalities are linguistic pre-processing, sentiment analysis, named entity recognition and lexicon lookup respectively. Outputs produced by each module are combined into feature vectors. The structure of feature vectors will vary across the KLOG questions. Finally, Weka, a machine learning workbench, was used for classification of feature vectors.ResultsThe precision of the system varied between 62.8 and 95.3%, whereas the recall varied from 58.3 to 87.6% across the 10 questions. The overall performance in terms of F–measure varied between 59.0 and 91.3% with an average of 74.4% and a standard deviation of 8.8.ConclusionsWe demonstrated the feasibility of mining open-ended patient questionnaires. By automatically mapping free text answers onto a Likert scale, we can effectively measure the progress of rehabilitation over time. In comparison to traditional closed-ended questionnaires, our approach offers much richer information that can be utilised to support clinical decision making. In conclusion, we demonstrated how text mining can be used to combine the benefits of qualitative and quantitative analysis of patient experiences.

Application Research of a Practical and New Intelligent Question Answering System

Intelligent question-answering system (IQAS) can help users quickly retrieve the answers to the required questions which is currently a popular research direction of NLP. This paper combines the research of relevant scholars and the analysis of their advantages and disadvantages, and an intelligent question-and-answer system is designed, which can automatically generate a question-and-answer knowledge base based on a given document. It can retrieve the knowledge base and recommend answers automatically, quickly and accurately according to the keywords provided by users. The system realized text semantic understanding and analysis, and its MRR value reached a satisfactory 0.7381. At the same time, the paper extends an automatic answer text generation technology based on Seq2Seq model, which can be a useful supplement to the traditional question-and-answer recommendation generation strategy.

Automatic Classification of Open-Ended Questions: Check-All-That-Apply Questions

Text data from open-ended questions in surveys are challenging to analyze and are often ignored. Open-ended questions are important though because they do not constrain respondents’ answers. Where open-ended questions are necessary, often human coders manually code answers. When data sets are large, it is impractical or too costly to manually code all answer texts. Instead, text answers can be converted into numerical variables, and a statistical/machine learning algorithm can be trained on a subset of manually coded data. This statistical model is then used to predict the codes of the remainder. We consider open-ended questions where the answers are coded into multiple labels (all-that-apply questions). For example, in the open-ended question in our Happy example respondents are explicitly told they may list multiple things that make them happy. Algorithms for multilabel data take into account the correlation among the answer codes and may therefore give better prediction results. For example, when giving examples of civil disobedience, respondents talking about “minor nonviolent offenses” were also likely to talk about “crimes.” We compare the performance of two different multilabel algorithms (random k-labelsets [RAKEL], classifier chains [CC]) to the default method of binary relevance (BR) which applies single-label algorithms to each code separately. Performance is evaluated on data from three open-ended questions (Happy, Civil Disobedience, and Immigrant). We found weak bivariate label correlations in the Happy data (90th percentile: 7.6%), and stronger bivariate label correlations in the Civil Disobedience (90th percentile: 17.2%) and Immigrant (90th percentile: 19.2%) data. For the data with stronger correlations, we found both multilabel methods performed substantially better than BR using 0/1 loss (“at least one label is incorrect”) and had little effect when using Hamming loss (average error). For data with weak label correlations, we found no difference in performance between multilabel methods and BR. We conclude that automatic classification of open-ended questions that allow multiple answers may benefit from using multilabel algorithms for 0/1 loss. The degree of correlations among the labels may be a useful prognostic tool.