Understanding the “Pathway” Towards a Searcher’s Learning Objective

Search systems are often used to support learning-oriented goals. This trend has given rise to the “search-as-learning” movement, which proposes that search systems should be designed to support learning. To this end, an important research question is: How does a searcher’s type of learning objective (LO) influence their trajectory (or pathway ) toward that objective? We report on a lab study (N = 36) in which participants gathered information to meet a specific type of LO. To characterize LOs and pathways , we leveraged Anderson and Krathwohl’s (A&K’s) taxonomy [ 3 ]. A&K’s taxonomy situates LOs at the intersection of two orthogonal dimensions: (1) cognitive process (CP) (remember, understand, apply, analyze, evaluate, and create) and (2) knowledge type (factual, conceptual, procedural, and metacognitive knowledge). Participants completed learning-oriented search tasks that varied along three CPs (apply, evaluate, and create) and three knowledge types (factual, conceptual, and procedural knowledge). A pathway is defined as a sequence of learning instances (e.g., subgoals) that were also each classified into cells from A&K’s taxonomy. Our study used a think-aloud protocol, and pathways were generated through a qualitative analysis of participants’ think-aloud comments and recorded screen activities. We investigate three research questions. First, in RQ1, we study the impact of the LO on pathway characteristics (e.g., pathway length). Second, in RQ2, we study the impact of the LO on the types of A&K cells traversed along the pathway. Third, in RQ3, we study common and uncommon transitions between A&K cells along pathways conditioned on the knowledge type of the objective. We discuss implications of our results for designing search systems to support learning.

Bilateral Personalized Dialogue Generation with Contrastive Learning

Generating personalized responses is one of the major challenges in natural human-robot interaction. Current researches in this field mainly focus on generating responses consistent with the robot’s pre-assigned persona, while ignoring the user’s persona. Such responses may be inappropriate or even offensive, which may lead to the bad user experience. Therefore, we propose a Bilateral Personalized Dialogue Generation (BPDG) method for dyadic conversation, which integrates user and robot personas into dialogue generation via designing a dynamic persona-aware fusion method. To bridge the gap between the learning objective function and evaluation metrics, the Conditional Mutual Information Maximum (CMIM) criterion is adopted with contrastive learning to select the proper response from the generated candidates. Moreover, a bilateral persona accuracy metric is designed to measure the degree of bilateral personalization. Experimental results demonstrate that, compared with several state-of-the-art methods, the final results of the proposed method are more personalized and consistent with bilateral personas in terms of both automatic and manual evaluations.

A personalized programming exercise recommendation algorithm based on knowledge structure tree

Personalized exercise recommendation is an important research project in the field of online learning, which can explore students’ strengths and weaknesses and tailor exercises for them. However, programming exercises differs from other disciplines or types of exercises due to the comprehensive of the exercises and the specificity of program debugging. In order to assist students in learning programming, this paper proposes a programming exercise recommendation algorithm based on knowledge structure tree (KSTER). Firstly, the algorithm provides a calculation method for quantifying students’ cognitive level to obtain their knowledge needs through individual learning-related data. Secondly, a knowledge structure tree is constructed based on the association relationship of knowledge points, and a learning objective prediction method is proposed by combining the knowledge needs and the knowledge structure tree to represent and update the learning objective. Finally, KSTER imports a matching operator that calculates cognitive level and exercise difficulty based on learning objectives, and makes top-η recommendation for exercises. Experiments show that the proposed algorithm significantly outperforms the other algorithms in both precision and recall. The comparison experiments with real-world data demonstrate that KSTER effectively improves students’ learning efficiency.

How to Implement Game-Based Learning in a Smart Classroom? A Model Based on a Systematic Literature Review and Delphi Method

Game-based learning (GBL) can allow learners to acquire and construct knowledge in a fun and focused learning atmosphere. A systematic literature review of 42 papers from 2010 to 2020 in this study showed that the current difficulties in implementing GBL in classrooms could be classified into the following categories: infrastructure, resources, theoretical guidance, teacher’s capabilities and acceptance of GBL. In order to solve the above problems, the study constructs a technology enhanced GBL model, from the four parts of learning objective, learning process, learning evaluation, and smart classroom. In addition, this study adopted the Delphi method, inviting a total of 29 scholars, experts, teachers and school managers to explore how to implement GBL in smart classrooms. Finally, the technology enhanced GBL model was validated and the utilization approaches were provided at the conclusion part.

Education on Blastic Plasmacytoid Dendritic Cell Neoplasm Significantly Impacts the Interdisciplinary Physician Team

Background: Blastic plasmacytoid dendritic cell neoplasm (BPDCN) is a rare, aggressive malignancy that originates from precursors of plasmacytoid dendritic cells. BPDCN is a difficult disease to diagnose and manage and it is often misdiagnosed or underreported. The literature widely supports the need for an interdisciplinary team of physicians with specialized expertise to care for patients with BPDCN, such as dermatologists, pathologists, hematologists/oncologists (hem/oncs), stem cell transplant physicians and others. Aim: The objective of this study was to determine if online education could improve the knowledge of the interdisciplinary physician team members about BPDCN as well as their skills and confidence diagnosing this rare malignancy. Methods: Dermatologists, pathologists, and hem/oncs participated in a series of 6 live continuing medical education (CME)-certified activities, after which the recorded content was posted online as a single online enduring CME-certified activity. Content for the CME activities was developed by a multidisciplinary group of BPDCN experts and was delivered through an approximately 1-hour lecture. Data presented here is for the online enduring activity only. Educational effect was assessed using a repeated-pair design with pre-/post-assessment. Three multiple choice questions assessed knowledge/skills, and 1 question rated on a Likert-type scale assessed confidence. A paired samples t-test was conducted for significance testing on overall average number of correct responses and for confidence rating, and a McNemar's test was conducted at the question and learning objective level (5% significance level, P <.05). Data were collected from December 10, 2020 to May 3, 2021. Results: There were 246 dermatologists, 302 pathologists, and 316 hem/oncs included in this analysis, for overall n=864. PRACTICE SETTING: Dermatologists: 57% community, 15% academic, 13% government, 15% other; Pathologists: 37% community, 37% other, 23% academic, 4% government; Hem/Oncs: 48% community, 31% academic, 14% government, 7% other.OVERALL RESULTS: Overall 46% of dermatologists, 42% of pathologists, and 48% of hem/oncs improved their knowledge/skills related to BPDCN (P <.001 for all), showing a relative increase in responses correct from pre- to post-CME of 67% for dermatologists, 38% for pathologists, and 45% for hem/oncs.CONFIDENCE: 50% of dermatologists, 50% of pathologists, and 49% of hem/oncs had a measurable increase in confidence (P <.001 for all), resulting in 30% of dermatologists, 31% of pathologists, and 36% of hem/oncs who were mostly or very confident diagnosing BPDCN post-CME (9%, 14%, 17% pre-CME, respectively).The Table shows the mean percentage of correct responses by learning objective and the question used to test each learning objective. 20%/54% of dermatologists, 22%/57% of pathologists, and 20%/55% of hem/oncs improved/reinforced their knowledge of the most common cutaneous manifestations of BPDCN and 26%, 22%, 25% need additional education, respectively. CME improved skills ordering tests to diagnose BPDCN, however 57% of dermatologists, 58% of pathologists, and 45% of hem/oncs demonstrate a need for additional education about stains that can aid in diagnosing BPDCN. Conclusions: This online CME-certified educational activity led to statistically significant improvements in the knowledge and skills of dermatologists, pathologists, and hem/oncs about BPDCN as well as their skills and confidence diagnosing this rare malignancy. The results indicate that unique educational methodologies which are available on-demand can be effective tools for advancing clinical decision making. Additional education is recommended on the topics of cutaneous manifestations of BPDCN and case-based education to improve skills diagnosing BPDCN. Acknowledgements: This CME activity was supported by an independent educational grant from Stemline Therapeutics, Inc. Reference: https://www.medscape.org/viewarticle/942245 Figure 1 Figure 1. Disclosures Stein: Amgen: Consultancy, Speakers Bureau; Celgene: Speakers Bureau; Stemline: Speakers Bureau. Sweet: Novartis: Honoraria, Membership on an entity's Board of Directors or advisory committees; Gilead: Membership on an entity's Board of Directors or advisory committees; AROG: Membership on an entity's Board of Directors or advisory committees; Astellas: Consultancy, Membership on an entity's Board of Directors or advisory committees; Bristol Meyers Squibb: Honoraria, Membership on an entity's Board of Directors or advisory committees. Guitart: Miragen, Kyowa Kirin: Consultancy; Galderma: Consultancy, Research Funding; Solygenix, Elorac, Nanostring: Research Funding. Pemmaraju: LFB Biotechnologies: Consultancy; Aptitude Health: Consultancy; Stemline Therapeutics, Inc.: Consultancy, Membership on an entity's Board of Directors or advisory committees, Other, Research Funding; Incyte: Consultancy; Daiichi Sankyo, Inc.: Other, Research Funding; Plexxicon: Other, Research Funding; Springer Science + Business Media: Other; Cellectis S.A. ADR: Other, Research Funding; CareDx, Inc.: Consultancy; Affymetrix: Consultancy, Research Funding; Roche Diagnostics: Consultancy; Novartis Pharmaceuticals: Consultancy, Other: Research Support, Research Funding; Blueprint Medicines: Consultancy; Celgene Corporation: Consultancy; DAVA Oncology: Consultancy; Sager Strong Foundation: Other; ASCO Leukemia Advisory Panel: Membership on an entity's Board of Directors or advisory committees; ASH Communications Committee: Membership on an entity's Board of Directors or advisory committees; MustangBio: Consultancy, Other; Abbvie Pharmaceuticals: Consultancy, Membership on an entity's Board of Directors or advisory committees, Other, Research Funding; Dan's House of Hope: Membership on an entity's Board of Directors or advisory committees; HemOnc Times/Oncology Times: Membership on an entity's Board of Directors or advisory committees; Samus: Other, Research Funding; Bristol-Myers Squibb Co.: Consultancy; Protagonist Therapeutics, Inc.: Consultancy; Clearview Healthcare Partners: Consultancy; ImmunoGen, Inc: Consultancy; Pacylex Pharmaceuticals: Consultancy. Poligone: Stemline, Helsinn, Kyowa Kirin: Consultancy; Soligenix, Miragen, Helsinn, Bioniz: Research Funding; Stemline, Therakos, Regeneron: Speakers Bureau.

Positive-Unlabeled Learning from Imbalanced Data

Positive-unlabeled (PU) learning deals with the binary classification problem when only positive (P) and unlabeled (U) data are available, without negative (N) data. Existing PU methods perform well on the balanced dataset. However, in real applications such as financial fraud detection or medical diagnosis, data are always imbalanced. It remains unclear whether existing PU methods can perform well on imbalanced data. In this paper, we explore this problem and propose a general learning objective for PU learning targeting specially at imbalanced data. By this general learning objective, state-of-the-art PU methods based on optimizing a consistent risk can be adapted to conquer the imbalance. We theoretically show that in expectation, optimizing our learning objective is equivalent to learning a classifier on the oversampled balanced data with both P and N data available, and further provide an estimation error bound. Finally, experimental results validate the effectiveness of our proposal compared to state-of-the-art PU methods.

Contrastive Model Invertion for Data-Free Knolwedge Distillation

Model inversion, whose goal is to recover training data from a pre-trained model, has been recently proved feasible. However, existing inversion methods usually suffer from the mode collapse problem, where the synthesized instances are highly similar to each other and thus show limited effectiveness for downstream tasks, such as knowledge distillation. In this paper, we propose Contrastive Model Inversion (CMI), where the data diversity is explicitly modeled as an optimizable objective, to alleviate the mode collapse issue. Our main observation is that, under the constraint of the same amount of data, higher data diversity usually indicates stronger instance discrimination. To this end, we introduce in CMI a contrastive learning objective that encourages the synthesizing instances to be distinguishable from the already synthesized ones in previous batches. Experiments of pre-trained models on CIFAR-10, CIFAR-100, and Tiny-ImageNet demonstrate that CMI not only generates more visually plausible instances than the state of the arts, but also achieves significantly superior performance when the generated data are used for knowledge distillation. Code is available at https://github.com/zju-vipa/DataFree.

Learning Class-Transductive Intent Representations for Zero-shot Intent Detection

Zero-shot intent detection (ZSID) aims to deal with the continuously emerging intents without annotated training data. However, existing ZSID systems suffer from two limitations: 1) They are not good at modeling the relationship between seen and unseen intents. 2) They cannot effectively recognize unseen intents under the generalized intent detection (GZSID) setting. A critical problem behind these limitations is that the representations of unseen intents cannot be learned in the training stage. To address this problem, we propose a novel framework that utilizes unseen class labels to learn Class-Transductive Intent Representations (CTIR). Specifically, we allow the model to predict unseen intents during training, with the corresponding label names serving as input utterances. On this basis, we introduce a multi-task learning objective, which encourages the model to learn the distinctions among intents, and a similarity scorer, which estimates the connections among intents more accurately. CTIR is easy to implement and can be integrated with existing ZSID and GZSID methods. Experiments on two real-world datasets show that CTIR brings considerable improvement to the baseline systems.

Virtual, Augmented and Mixed Realities Techniques Applied to University Research and Teaching in the field of Physics.

The incorporation of the new audiovisual technology to the closest social field allows you to use your familiarity and easy access to the research and teaching disciplines of learning also in the university. In particular, it is of utmost importance for the Teaching Innovation Project (TIP) that is addressed in this document: “Virtual, Augmented and Mixed Realities Techniques Applied to University Research and Teaching in the field of Physics”. It has great advantages for training / learning especially for the last generations, so familiar with all kinds of audiovisual technology. Obviously, introductory complements to the field of specific competencies are needed so that the fundamental training and meaningful learning objective comes to fruition. For this, it is necessary to have a good information base on Physics treated through Virtual, Augmented and Mixed Realities Techniques, in order to be able to select the appropriate information and level in each of the TIP stages. This base is susceptible to evolution and improvement if sufficient tools and knowledge are available. In addition, it will be possible to generate new procedures based on the strengths and weaknesses appreciated in this TIP.