Read, Attend, and Exclude: Multi-Choice Reading Comprehension by Mimicking Human Reasoning Process

Author(s):  
Chenbin Zhang ◽  
Congjian Luo ◽  
Junyu Lu ◽  
Ao Liu ◽  
Bing Bai ◽  
...  
Author(s):  
Zeyun Tang ◽  
Yongliang Shen ◽  
Xinyin Ma ◽  
Wei Xu ◽  
Jiale Yu ◽  
...  

Multi-hop reading comprehension across multiple documents attracts much attentions recently. In this paper, we propose a novel approach to tackle this multi-hop reading comprehension problem. Inspired by the human reasoning processing, we introduce a path-based graph with reasoning paths which extracted from supporting documents. The path-based graph can combine both the idea of the graph-based and path-based approaches, so it is better for multi-hop reasoning. Meanwhile, we propose Gated-GCN to accumulate evidences on the path-based graph, which contains a new question-aware gating mechanism to regulate the usefulness of information propagating across documents and add question information during reasoning. We evaluate our approach on WikiHop dataset, and our approach achieves the the-state-of-art accuracy against previous published approaches. Especially, our ensemble model surpasses the human performance by 4.2%.


2014 ◽  
Vol 42 (4) ◽  
pp. 635-643
Author(s):  
Le Zhao ◽  
Xu Chen

Human reasoning is often biased by heuristic thinking, but the nature of heuristic bias is yet to be exactly explained. As the focus of most relevant experiments on heuristic bias in reasoning has been exclusively on adults, we conducted a study with children aged between 7 and 16 years (N = 129). Our aim was to identify which element in the reasoning process fails and thus underlies the heuristic bias in children's reasoning in the personal domain. We explored the probability of judging violation among this group of children in reasoning tasks involving personal domain rules and by measuring their lexical decision times. The results showed that the heuristic bias in children's reasoning in the personal domain was caused by monitoring failure among those who were of junior school age and by inhibition failure among those who were of middle school age. As most of the children of senior school age were able to complete the inhibition process, they had less heuristic bias.


2020 ◽  
Vol 43 ◽  
Author(s):  
Ian Robertson

Abstract Osiurak and Reynaud (O&R) claim that research into the origin of cumulative technological culture has been too focused on social cognition and has consequently neglected the importance of uniquely human reasoning capacities. This commentary raises two interrelated theoretical concerns about O&R's notion of technical-reasoning capacities, and suggests how these concerns might be met.


2019 ◽  
Vol 50 (4) ◽  
pp. 562-578 ◽  
Author(s):  
Dawna Duff

Purpose Vocabulary intervention can improve comprehension of texts containing taught words, but it is unclear if all middle school readers get this benefit. This study tests 2 hypotheses about variables that predict response to vocabulary treatment on text comprehension: gains in vocabulary knowledge due to treatment and pretreatment reading comprehension scores. Method Students in Grade 6 ( N = 23) completed a 5-session intervention based on robust vocabulary instruction (RVI). Knowledge of the semantics of taught words was measured pre- and posttreatment. Participants then read 2 matched texts, 1 containing taught words (treated) and 1 not (untreated). Treated texts and taught word lists were counterbalanced across participants. The difference between text comprehension scores in treated and untreated conditions was taken as a measure of the effect of RVI on text comprehension. Results RVI resulted in significant gains in knowledge of taught words ( d RM = 2.26) and text comprehension ( d RM = 0.31). The extent of gains in vocabulary knowledge after vocabulary treatment did not predict the effect of RVI on comprehension of texts. However, untreated reading comprehension scores moderated the effect of the vocabulary treatment on text comprehension: Lower reading comprehension was associated with greater gains in text comprehension. Readers with comprehension scores below the mean experienced large gains in comprehension, but those with average/above average reading comprehension scores did not. Conclusion Vocabulary instruction had a larger effect on text comprehension for readers in Grade 6 who had lower untreated reading comprehension scores. In contrast, the amount that children learned about taught vocabulary did not predict the effect of vocabulary instruction on text comprehension. This has implications for the identification of 6th-grade students who would benefit from classroom instruction or clinical intervention targeting vocabulary knowledge.


2009 ◽  
Vol 10 (3) ◽  
pp. 82-89
Author(s):  
Janet L. Proly ◽  
Jessica Rivers ◽  
Jamie Schwartz

Abstract Graphic organizers are a research based strategy used for facilitating the reading comprehension of expository text. This strategy will be defined and the evolution and supporting evidence for the use of graphic organizers will be discussed. Various types of graphic organizers and resources for SLPs and other educators will also be discussed.


2013 ◽  
Vol 27 (4) ◽  
pp. 283-293 ◽  
Author(s):  
Lars Behrmann ◽  
Elmar Souvignier

Single studies suggest that the effectiveness of certain instructional activities depends on teachers' judgment accuracy. However, sufficient empirical data is still lacking. In this longitudinal study (N = 75 teachers and 1,865 students), we assessed if the effectiveness of teacher feedback was moderated by judgment accuracy in a standardized reading program. For the purpose of a discriminant validation, moderating effects of teachers' judgment accuracy on their classroom management skills were examined. As expected, multilevel analyses revealed larger reading comprehension gains when teachers provided students with a high number of feedbacks and simultaneously demonstrated high judgment accuracy. Neither interactions nor main effects were found for classroom management skills on reading comprehension. Moreover, no significant interactions with judgment accuracy but main effects were found for both feedback and classroom management skills concerning reading strategy knowledge gains. The implications of the results are discussed.


1983 ◽  
Vol 28 (1) ◽  
pp. 31-31
Author(s):  
John B. Carroll

Sign in / Sign up

Export Citation Format

Share Document