Appearance
Glossary
Each term has one definition. Concept pages expand on them.
Annotation
A label you attach to a time range of a video. It has a type, a start time, an end time, optional content, one speaker, and the coder who made it.
Annotation type
A kind of thing you annotate, such as "Gaze" or "Gesture". Types have a colour and optional sub-types.
Category
A group of annotation types, such as EMBODIED MODES. Coverage and some charts report one figure per category.
Track
One horizontal row in the timeline. Each track shows the annotations of one annotation type. A track either counts annotations or measures how long they last.
Sub-type
A more specific value inside an annotation type, such as "Gaze → at student".
Library
The set of annotation types a project uses. Saved libraries can be reused in other projects.
Project
A folder for videos, their annotations, and the library they share.
Video
A media file uploaded to GRAPE MARS. A video can belong to more than one project.
Speaker
A person in the video. Each annotation belongs to one speaker.
Coder
A person who makes annotations. GRAPE MARS records which user made each annotation.
Transcript
The written text of what is said in the video, split into segments with times.
Segment
One line of a transcript, with a start time and end time.
Word-level timing
Timing for every word in a transcript, not only for each segment. AI transcription produces it.
Subtitles
A lane in the timeline that shows the transcript segments as blocks under the video.
Lemma
The dictionary form of a word. "running" and "ran" have the lemma "run".
Part of speech
The word class of a word: noun, verb, adjective, and so on.
Named entity
A word or phrase that names something: a person, a place, an organisation, a date.
Tagging
The automatic step that reads a transcript and records, for every word, its lemma, its part of speech, and whether it names something.
Concordance
A search that shows every occurrence of a word or annotation with the words around it. Researchers also call this a KWIC view (key word in context).
Precision (word or segment)
How exactly a concordance result is placed in time. "Word" means the exact word. "Segment" means the transcript line that contains it.
Cohen's kappa
A score from -1 to 1 that shows how much 2 coders agree, after removing agreement that would happen by luck. A kappa of 0.8 means substantial agreement.
Co-occurrence
2 annotations of different types that overlap in time, or fall within a short window of each other.
Coverage
The merged share of a video's duration that has at least one annotation, shown for each category.
Density
How many annotations are active in each moment of a video.
Fragment
A time range inside a video, used to limit an analysis.
Module
One analysis view inside a workspace, such as a chart or a table.
Workspace
A saved page of analysis modules for one project.
Scope
The videos, time range, tracks, coders, and speakers a workspace or module looks at.
Role
What a person can do in a project: owner, editor, or viewer.
Autosave
GRAPE MARS saves each change as you make it. There is no save button in the annotator.
AI key
A personal key from Mistral AI or Google Gemini that lets GRAPE MARS use their AI on your behalf.
Super-admin
An administrator who can manage users, projects, and platform settings.