An official website of the United States government
Online lecture videos are increasingly important e-learning materials for students. Automated content extraction from lecture videos facilitates information retrieval applications that improve access to the lecture material. A significant number of lecture videos include the speaker in the image. Speakers perform various semantically meaningful actions during the process of teaching. Among all the movements of the speaker, key actions such as writing or erasing potentially indicate important features directly related to the lecture content. In this paper, we present a methodology for lecture video content extraction using the speaker actions. Each lecture video is divided into small temporal units called action segments. Using a pose estimator, body and hands skeleton data are extracted and used to compute motion-based features describing each action segment. Then, the dominant speaker action of each of these segments is classified using Random forests and the motion-based features. With the temporal and spatial range of these actions, we implement an alternative way to draw key-frames of handwritten content from the video. In addition, for our fixed camera videos, we also use the skeleton data to compute a mask of the speaker writing locations for the subtraction of the background noise from the binarized key-frames. Our method has been tested on a publicly available lecture video dataset, and it shows reasonable recall and precision results, with a very good compression ratio which is better than previous methods based on content analysis.
more »
« less
This content will become publicly available on May 1, 2026
Interpolated Retrieval of Relevant Material, Not Irrelevant Material, Enhances New Learning of a Video Lecture In-Person and Online
Interpolated retrieval enhances the learning of new information—a finding known as the forward testing effect. The context change account suggests that learning benefits are due to a shift in internal context, which can be triggered through the retrieval of either content-relevant or content-irrelevant information. In two experiments, we examined whether interpolated episodic, autobiographical, and semantic retrieval would enhance new learning of a video lecture, compared to interpolated review. Participants watched a STEM topic lecture divided into three ~5 min segments and completed their assigned interpolated activity after the first two segments. Across both a laboratory (Experiment 1, N = 249) and online setting (Experiment 2, N = 246), only episodic retrieval enhanced the learning of new material; autobiographical and semantic retrieval (content-irrelevant) did not improve new learning. Critically, we introduced a measure of context change to determine whether the level of engagement in these interpolated activities predicted recall. Engagement correlated with criterial test performance when controlling for effort (seriousness). Our results support a multi-factor explanation for the forward testing effect, providing evidence for both the context change and strategy change accounts, although we emphasize that support for context change should be interpreted with caution.
more »
« less
- Award ID(s):
- 2017333
- PAR ID:
- 10603995
- Publisher / Repository:
- MDPI
- Date Published:
- Journal Name:
- Behavioral Sciences
- Volume:
- 15
- Issue:
- 5
- ISSN:
- 2076-328X
- Page Range / eLocation ID:
- 668
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract BackgroundUsing simulations in science instruction can help make abstract topics more concrete and boost students' understanding. AimsThe current research examined whether using a simulation as an exploratory learning activity before an accompanying lecture has additional learning and motivational benefits compared to a more common lecture‐then‐simulation approach. SamplesParticipants (Experiment 1,N = 168; Experiment 2,N = 357) were undergraduate students in several sections of a first‐year chemistry course. MethodsStudents were randomly assigned to explore a simulation on atomic structure either before a lecture (explore‐first condition) or after the lecture (instruct‐first condition). In Experiment 1, the simulation activity time was limited (15 min) and the activity varied in whether self‐explanation (‘why’) prompts were included. In Experiment 2, the activity time was lengthened (20 min), and only ‘why’ prompts were used. After the activity and lecture, students completed a survey and posttest. ResultsIn Experiment 1, students in the explore‐first condition scored lower on posttest conceptual knowledge scores and reported lower curiosity compared to students in the instruct‐first condition. Scores for basic facts and transfer knowledge, and self‐reported situational interest, self‐efficacy, and competence, were equal between conditions. No effects of prompt condition were found. In Experiment 2, with longer activity time, the results reversed. Students in the explore‐first condition scored equally on basic facts and higher on conceptual knowledge and transfer measures, while also reporting higher curiosity, situational interest, self‐efficacy, competence, and cognitive engagement. ConclusionWhen properly designed, placing simulations before—rather than after—lecture can deepen learning, motivation, and competence.more » « less
-
The current study assesses whether varying the encoding context of a repeated event is a potential strategy to improve recognition memory across retrieval contexts. Context variability, also known as encoding variability, has historically been investigated primarily using recall and cued recall tasks, with the consensus being that encoding variability is not necessarily beneficial for episodic retrieval. However, recent studies (see text) suggest that test type may determine the strategy’s effectiveness. Aligned with these recent findings, we found consistent benefits to simple item recognition when a word was studied in more variable contexts compared to less variable contexts across four experiments. This main effect of context variability occurred when crossed with a manipulation of repetition spacing and when crossed with a manipulation of encoding-retrieval context match. Variation in encoding contexts beyond the future retrieval context led to better item recognition than repeated study exposures within the future retrieval context. We argue that the current study and other recent findings indicate a need to re-evaluate the historical consensus on encoding variability as a beneficial strategy for learning.more » « less
-
Context contributes to multiple aspects of human episodic memory including segmentation and retrieval. The present studies tested if, in adult male and female mice, context influences the encoding of odors encountered in a single unsupervised sampling session of the type used for the routine acquisition of episodic memories. The three paradigms used differed in complexity (single vs. multiple odor cues) and period from sampling to testing. Results show that males consistently encode odors in a context-dependent manner: the mice discriminated novel from previously sampled cues when tested in the chamber of initial cue sampling but not in a distinct yet familiar chamber. This was independent of the interval between cue encounters or the latency from initial sampling to testing. In contrast, female mice acquired both single cues and the elements of multi-cue episodes, but recall of that information was dependent upon the surrounding context only when the cues were presented serially. These results extend the list of episodic memory features expressed by rodents and also introduce a striking and unexpected sex difference in context effects.more » « less
-
null (Ed.)We address the problem of ad hoc table retrieval via a new neural architecture that incorporates both semantic and relevance matching. Understanding the connection between the structured form of a table and query tokens is an important yet neglected problem in information retrieval. We use a learning- to-rank approach to train a system to capture semantic and relevance signals within interactions between the structured form of candidate tables and query tokens. Convolutional filters that extract contextual features from query/table interactions are combined with a feature vector based on the distributions of term similarity between queries and tables. We propose using row and column summaries to incorporate table content into our new neural model. We evaluate our approach using two datasets, and we demonstrate substantial improvements in terms of retrieval metrics over state-of-the-art methods in table retrieval and document retrieval, and neural architectures from sentence, document, and table type classification adapted to the table retrieval task. Our ablation study supports the importance of both semantic and relevance matching in the table retrieval.more » « less
