The Caliper Framework Fills a Need for Classroom Video Analytics

今年4月, “下一代数码学习环境”(NGDLE), the influential white paper from Educause Learning Initiative, 庆祝其出版四周年. To recap: It calls for standardized modularity and interoperability in the next generation of educational technology tools.

The transition to the NGDLE era has thus far been evolutionary rather than revolutionary; vendors of established, feature-rich ed tech products have adopted interoperability standards where it benefits them, and startup ed tech providers have formed to fill feature niches where interoperability standards exist to enable them.

为教育部门提供视频服务的供应商, NGDLE运动几乎不构成生存威胁. Video services provide specialized web applications which need resources that scale very differently from LMSs, so they have always filled feature niches that don’t make sense for LMS vendors to provide internally. The adoption of interoperability standards has actually benefitted video platforms: Consider the clever ways that many have used the IMS Global Learning Consortium’s Learning Tools Interoperability (LTI) standard for course-registration-based access control to videos in support of FERPA (Family Educational Rights and Privacy Act) and of TEACH Act (Technology, 教育和版权协调)合规性.

随着运动走向更加模块化, loosely coupled online classroom comes the problem of fragmented analytics. LTI activities and video services might have gee-whiz internal views for teachers to study how students used the product, but the LTI standard does not support reporting usage data to the LMS, 除了成绩. 因此, teachers lack a straightforward means to associate granular student behaviors in video services or LTI applications with any other behaviors or with the broader learning trends in the course. According to the canonical course activity logs provided to teachers from the LMS, student activities taking place on third-party platforms don’t exist. 没有学生在课程中表现的综合数据, 视频使用数据几乎只是一个噱头.

进入卡尺, the IMS framework whereby learning activities—be they run within the LMS, LTI应用, or some other compliant web application—implement sensors that emit log reports as structured JSON documents to a central event store which handles the cross-platform reporting and analytic aggregation for teachers. Caliper was designed primarily to allow schools to study student behaviors at big-data scales in support of data-driven approaches to improve instruction, and so defines a well-considered event ontology to standardize the events being logged.

Caliper定义了一个MediaEvent with actions that cover playback events like start, end, pause, and seek. It also supports dynamic streaming actions such as when the resolution changes, as well as purely client-side events like when the user changes the volume or playback rate, 或切换字幕, 全屏, 还有音频静音. The framework permits unconstrained elaboration on the reported action within an “extensions” property. Video platforms that support in-video quizzing can also emit AssessmentItemEvents for each question presented to students.

At least some subset of these actions is likely being logged by any video platform that serves education: Implementing a compliant sensor to emit them to a system-administrator-configured event store endpoint is a tractable feature addition, especially considering that IMS Global supports actively maintained reference implementations for 六种常用的编程语言.

Caliper events can be emitted by many LMSs, including Blackboard, Canvas, and Moodle. Blackboard’s cloud-based LMS offering even serves as a Caliper Event store itself, so third-party services can emit events directly to the LMS to be integrated with native LMS events for teachers to review as the standard intended. Of note: Blackboard’s documentation for configuring the LMS to receive Caliper events from another service uses a video platform, 创作, 作为它的范例.

[本文发表于2019年4月/ 5月号。 流媒体杂志 即“聚合学习分析”]