Glossary¶
A short glossary of terms used in this document, explained in plain words.
| Term | In plain words |
|---|---|
| ControlAI | A system that automatically analyzes every class session from camera recordings and shows the result in the admin panel. |
| Admin panel | A website university staff log into under their role to see reports, metrics, and management tools. The product's main (and only) interface. |
| Class session (para) | A standard university class, typically 80 minutes (2 academic hours). The unit of analysis in ControlAI: one class session → one report. |
| Group (guruh) | An academic group with a year of study, major, and language of instruction (Uzbek / Russian / mixed). |
| Class type | Lecture (ma'ruza), seminar, practical class (amaliy), or lab session. Known from the schedule and affects how the session's metrics should be read. |
| Department | A faculty subdivision that groups instructors by subject. In analytics — a comparison level: department vs. department, and instructor ratings within a department. |
| Faculty / dean's office | A university's major subdivision and its leadership (the dean). A university has several faculties — the panel shows and compares each one. |
| Rectorate | The university's leadership (rector, vice-rectors) — sees the consolidated picture across the whole university. |
| Vice-rector for academic affairs | A member of the rectorate responsible for the academic process. In ControlAI — the main user of university-level summaries: reviews them weekly and prepares a summary for the rectorate meeting. |
| Quality-control office (ichki ta'lim sifati nazorati bo'limi) | The university service that checks teaching quality. In ControlAI — the main "reader" of reports and reviewer of ratings. |
| Academic affairs office (o'quv bo'limi) | The university service responsible for the class schedule and organizing the academic process. |
| Speech recognition | Turning what was said during a class session into text. |
| "Who spoke when" | Splitting the recording by speaker: where the instructor spoke, where the students spoke (and how much). |
| Transcript | A text record of the class session — who said what, with timestamps. |
| Metric | Any measurable indicator of a class session: talk-time balance, number of questions, silence, and so on. |
| Talk-time balance | How much time the instructor spoke versus how much time the students spoke. |
| Target-language share | For language courses: what portion of the class was conducted in the target language versus the native language. |
| "Lesson DNA" | A color-coded timeline of the class session showing how time was split (instructor / students / questions / silence / group work). |
| Rubric (lesson observation form) | A list of quality criteria the university uses to evaluate class sessions. ControlAI fills it in automatically. |
| AI-nazoratchi / methodology compliance | The part of the product that checks a class session against the rubric and the university's methodological requirements — instead of a manual reviewer. |
| Open (demonstration) lesson (ochiq dars) | The traditional review format: a committee sits in on the class session with an observation form — usually once or twice a year. With ControlAI, that form is filled out for every class session. |
| Topic coverage | What in the course syllabus has already been covered, what's partially covered, and what hasn't been covered yet. |
| Instructor rating | An overall score for an instructor's work (1–10), built from factors: talk-time balance, questions, language, punctuality. Internal analytics for development — not a replacement for official certification. |
| Rating factor | One of the components that make up the rating, each with its own weight and explanation. |
| Skills radar | A 5-axis "star" chart showing an instructor's strengths and weaknesses (the four rating factors plus engagement). |
| Coaching buckets | Automatically grouping instructors into "recognize / support / watch". |
| Churn risk | A panel section (named as in the interface); for a university this means enrollment risk: which group is accumulating absences and academic debts, and where the risk of expulsion is higher. |
| Attendance / headcount | How many students were in the classroom — calculated from video. Roadmap stage 2, not a current feature. |
| Punctuality | How closely the class session started and ended on schedule. |
| "Lesson not held" (no_show) | A class that effectively didn't happen (the room was empty or the group didn't show up). Determined while processing the recording; it isn't analyzed and doesn't count toward the billed volume under the contract. |
| Operations (ops inbox) | A panel section listing current issues (camera offline, no audio, a failure, etc.) with an action for each. |
| Coverage (SLO) | A metric showing what share of class sessions were recorded and processed on time. |
| Video recorder (NVR/DVR) | The university's device that cameras record to; ControlAI pulls the recordings from there. It's already in place at a typical university — nothing new is installed in the classroom. |
| Voice ID (voice sample) | A voice sample that helps the system more accurately tell the instructor's speech apart from the students'. Only the instructor records it, and only voluntarily; the system doesn't identify individual students by voice. |
| HEMIS | The government information system for managing higher-education institutions (schedule, enrollment). Integration with it is roadmap stage 3; until then, the class schedule is entered in the panel. |
| Credit-module system | The model currently used to organize the academic process at Uzbekistan's universities; ControlAI's class schedule is built by semester within this system. |
| Dashboard | The main summary screen with key indicators and a "Needs attention" feed. |
| Consent | The instructor's permission to record and analyze their classes; for students (who are adults), transparently informing them is enough. Part of onboarding. |
The main point: this glossary is a cheat sheet for the other chapters. If you come across an unfamiliar term somewhere, it's most likely explained here in plain words.
← Back to the table of contents