The following sections detail how edX stores student state data internally, and is useful for developers and researchers who are examining database exports. This information includes demographic information collected at signup, course enrollment, course progress, and certificate status.
The following sections detail how edX stores stateful data for students internally, and is useful for developers and researchers who are examining database exports. This data includes:
* :ref:`User_Data`
* :ref:`Courseware_Progress`
* :ref:`Certificates`
Conventions to keep in mind:
* We currently use MySQL 5.1 with InnoDB tables
* EdX uses MySQL 5.1 relational database system with InnoDb storage engine.
* All strings are stored as UTF-8.
* All datetimes are stored as UTC.
* Tables that are built into the Django framework are not documented here unless we use them in unconventional ways.
* All datetimes are stored as UTC (Coordinated Universal Time).
.. note::
EdX also uses the Django Python Web framework. Tables that are built into the Django Web framework are documented here only if they are used in unconventional ways.
Descriptions of the tables and columns that store student data follow, first in summary form with field types and constraints, and then with a detailed explanation of each column.
********************
MySQL Terminology
********************
All of our tables will be described below, first in summary form with field types and constraints, and then with a detailed explanation of each field. For those not familiar with the MySQL schema terminology in the table summaries:
The summary information provided about the SQL table columns uses the following MySQL schema terminology.
`Type`
This is the kind of data it is, along with the size of the field. When a numeric field has a length specified, it just means that's how many digits we want displayed -- it has no affect on the number of bytes used.
========
Type
========
The kind of data and the size of the field. When a numeric field has a length specified, the length indicates how many digits display but does not affect the number of bytes used.
.. list-table::
:widths: 10 80
:widths: 25 65
:header-rows: 1
* - Value
- Meaning
* - `int`
- Description
* - int
- 4 byte integer.
* - `smallint`
* - smallint
- 2 byte integer, sometimes used for enumerated values.
* - `tinyint`
- 1 byte integer, but usually just used to indicate a boolean field with 0 = False and 1 = True.
* - `varchar`
- String, typically short and indexable. The length is the number of chars, not bytes (so unicode friendly).
* - `longtext`
* - tinyint
- 1 byte integer, usually used to indicate a Boolean with 0 = False and 1 = True.
* - varchar
- String, typically short and indexable. The length is the number of chars, not bytes, to support multi-byte character sets.
* - longtext
- A long block of text, usually not indexed.
* - `date`
* - date
- Date
* - `datetime`
* - datetime
- Datetime in UTC, precision in seconds.
`Null`
========
Null
========
.. list-table::
:widths: 10 80
:widths: 25 70
:header-rows: 1
* - Value
- Meaning
* - `YES`
- `NULL` values are allowed
* - `NO`
- `NULL` values are not allowed
- Description
* - YES
- NULL values are allowed.
* - NO
- NULL values are not allowed.
.. note::
Django often just places blank strings instead of NULL when it wants to indicate a text value is optional. This is more meaningful for numeric and date fields.
.. note::
Django often just places blank strings instead of NULL when it wants to indicate a text value is optional. This is used more meaningful for numeric and date fields.
========
Key
========
`Key`
.. list-table::
:widths: 10 80
:widths: 25 65
:header-rows: 1
* - Value
- Meaning
* - `PRI`
- Primary key for the table, usually named `id`, unique
* - `UNI`
- Description
* - PRI
- Primary key for the table, usually named ``id``, unique.
* - UNI
- Unique
* - `MUL`
- Indexed for fast lookup, but the same value can appear multiple times. A Unique index that allows `NULL` can also show up as `MUL`.
* - MUL
- Indexed for fast lookup, but the same value can appear multiple times. A unique index that allows NULL can also show up as MUL.
.. _User_Data:
****************
User Information
User Data
****************
`auth_user`
===========
The `auth_user` table is built into the Django web framework that we use. It holds generic information necessary for basic login and permissions information. It has the following fields::
Primary key, and the value typically used in URLs that reference the user. A user has the same value for `id` here as they do in the MongoDB database's users collection. Foreign keys referencing `auth_user.id` will often be named `user_id`, but are sometimes named `student_id`.
Primary key, and the value typically used in URLs that reference the user. A user has the same value for ``id`` here as they do in the MongoDB database's users collection. Foreign keys referencing ``auth_user.id`` will often be named ``user_id``, but are sometimes named ``student_id``.
`username`
----------
The unique username for a user in our system. It may contain alphanumeric, _, @, +, . and - characters. The username is the only information that the students give about themselves that we currently expose to other students. We have never allowed people to change their usernames so far, but that's not something we guarantee going forward.
username
----------
The unique username for a user in our system. It can contain alphanumerics and the special characters shown within the brackets: [ _ @ + - . ]. The username is the only user-provided information that other users can currently see. EdX has never allowed users to change usernames, but may do so in the future.
`first_name`
------------
.. note::
Not used; we store a user's full name in `auth_userprofile.name` instead.
first_name
------------
Not used; a user's full name is stored in ``auth_userprofile.name`` instead.
`last_name`
-----------
.. note::
Not used; we store a user's full name in `auth_userprofile.name` instead.
last_name
-----------
Not used; a user's full name is stored in ``auth_userprofile.name`` instead.
`email`
-------
Their email address. While Django by default makes this optional, we make it required, since it's the primary mechanism through which people log in. Must be unique to each user. Never shown to other users.
email
-------
The user's email address, which is the primary mechanism users use to log in. This value is optional by default in Django, but is required by edX. This value must be unique to each user and is never shown to other users.
`password`
----------
password
----------
A hashed version of the user's password. Depending on when the password was last set, this will either be a SHA1 hash or PBKDF2 with SHA256 (Django 1.3 uses the former and 1.4 the latter).
`is_staff`
----------
This value is `1` if the user is a staff member *of edX* with corresponding elevated privileges that cut across courses. It does not indicate that the person is a member of the course staff for any given course. Generally, users with this flag set to 1 are either edX program managers responsible for course delivery, or edX developers who need access for testing and debugging purposes. People who have `is_staff = 1` get instructor privileges on all courses, along with having additional debug information show up in the instructor tab.
is_staff
----------
Most users have a 0 for this field. Set to 1 if the user is a staff member of **edX**, with corresponding elevated privileges that cut across courses. It does not indicate that the person is a member of the course staff for any given course.
Note that this designation has no bearing with a user's role in the forums, and confers no elevated privileges there.
Generally, users with this flag set to 1 are either edX program managers responsible for course delivery, or edX developers who need access for testing and debugging purposes. Users who have ``is_staff`` = 1 have instructor privileges on all courses and can see additional debug information on the Instructor tab.
Most users have a `0` for this value.
.. note::
This designation has no bearing on a user's role in the discussion forums, and confers no elevated privileges there.
`is_active`
-----------
This value is `1` if the user has clicked on the activation link that was sent to them when they created their account, and `0` otherwise. Users who have `is_active = 0` generally cannot log into the system. However, when users first create their account, they are automatically logged in even though they are not active. This is to let them experience the site immediately without having to check their email. They just get a little banner at the top of their dashboard reminding them to check their email and activate their account when they have time. If they log out, they won't be able to log back in again until they've activated. However, because our sessions last a long time, it is theoretically possible for someone to use the site as a student for days without being "active".
is_active
-----------
This value is 1 if the user has clicked on the activation link that was sent to them when they created their account, and 0 otherwise.
Once `is_active` is set to `1`, the only circumstance where it would be set back to `0` would be if we decide to ban the user (which is very rare, manual operation).
Users who have ``is_active`` = 0 generally cannot log into the system. However, when users first create an account, they are automatically logged in even though they have not yet activated the account. This is to let them experience the site immediately without having to check their email. A message displays on the dashboard to remind users to check their email and activate their accounts when they have time. When they log out, they cannot log back in again until activation is complete. However, because edX sessions last a long time, it is possible for someone to use the site as a student for days without being "active".
`is_superuser`
Once ``is_active`` is set to 1, it is *only* set back to 0 if the user is banned (which is a very rare, manual operation).
--------------
is_superuser
--------------
Value is `1` if the user has admin privileges. Only the earliest developers of the system have this set to `1`, and it's no longer really used in the codebase. Set to 0 for almost everybody.
Controls access to django_admin views: set to 1 (true) only for site admins. 0 for almost everybody.
`last_login`
**History**: Only the earliest developers of the system have this set to 1, and it is no longer really used in the codebase.
------------
last_login
------------
A datetime of the user's last login. Should not be used as a proxy for activity, since people can use the site all the time and go days between logging in and out.
`date_joined`
-------------
Date that the account was created (NOT when it was activated).
date_joined
-------------
Date that the account was created.
`(obsolete fields)`
.. note::
This is not the date that the user activated the account.
-------------------
Obsolete columns
-------------------
All the following fields were added by an application called Askbot, a discussion forum package that is no longer part of the system:
* `status`
* `email_key`
* `avatar_type`
* `country`
* `show_country`
* `date_of_birth`
* `interesting_tags`
* `ignored_tags`
* `email_tag_filter_strategy`
* `display_tag_filter_strategy`
* `consecutive_days_visit_count`
Only users who were part of the prototype 6.002x course run in the Spring of 2012 would have any information in these fields. Even with those users, most of this information was never collected. Only the fields that are automatically generated have any values in them, such as tag settings.
These fields are completely unrelated to the discussion forums we currently use, and will eventually be dropped from this table.
`auth_userprofile`
==================
The `auth_userprofile` table is mostly used to store user demographic information collected during the signup process. We also use it to store certain additional metadata relating to certificates. Every row in this table corresponds to one row in `auth_user`::
There is an important split in demographic data gathered for the students who signed up during the MITx prototype phase in the spring of 2012, and those that signed up afterwards.
`id`
All of the following columns were added by an application called Askbot, a discussion forum package that is no longer part of the system.
* status
* email_key
* avatar_typ
* country
* show_country
* date_of_birth
* interesting_tags
* ignored_tags
* email_tag_filter_strategy
* display_tag_filter_strategy
* consecutive_days_visit_count
Only users who were part of the prototype 6.002x course run in the Spring of 2012 have any information in these columns. Even for those users, most of this information was never collected. Only the columns with values that are automatically generated have any values in them, such as the tag-related columns.
These columns are unrelated to the discussion forums that edX currently uses, and will eventually be dropped from this table.
.. _auth_userprofile:
======================================
Columns in the auth_userprofile Table
======================================
The ``auth_userprofile`` table stores user demographic data collected during the student registration process.
Every row in this table corresponds to one row in ``auth_user``.
A sample of the heading row and a data row in the ``auth_userprofile`` table follow.
.. code-block:: json
id user_id name language location meta courseware gender mailing_address
**History**: The organization of this table was different for the students who signed up for the MITx prototype phase in the spring of 2012, than for those who signed up afterwards. The column descriptions that follow detail the differences in the demographic data gathered.
----
id
----
Primary key, not referenced anywhere else.
`user_id`
---------
A foreign key that maps to `auth_user.id`.
user_id
---------
A foreign key that maps to ``auth_user.id``.
`name`
------
String for a user's full name. We make no constraints on language or breakdown into first/last name. The names are never shown to other students. Foreign students usually enter a romanized version of their names, but not always.
name
------
String for a user's full name. EdX makes no constraints on language or breakdown into first/last name. The names are never shown to other students. International students usually enter a romanized version of their names, but not always. Name changes are permitted, and the previous name is logged in the ``meta`` field.
It used to be our policy to require manual approval of name changes to guard the integrity of the certificates. Students would submit a name change request and someone from the team would approve or reject as appropriate. Later, we decided to allow the name changes to take place automatically, but to log previous names in the `meta` field.
**History**: A former edX policy required manual approval of name changes to guard the integrity of the certificates. Students would submit a name change request, and an edX team member would approve or reject the request.
`language`
----------
User's preferred language, asked during the sign up process for the 6.002x prototype course given in the Spring of 2012. This information stopped being collected after the transition from MITx to edX happened, but we never removed the values from our first group of students. Sometimes written in those languages.
language
----------
No longer used.
**History**: User's preferred language, asked during the sign up process for the 6.002x prototype course given in the Spring of 2012. Sometimes written in those languages. EdX stopped collecting this data after MITx transitioned to edX, but never removed the values for the first group of students.
`location`
----------
User's location, asked during the sign up process for the 6.002x prototype course given in the Spring of 2012. We weren't specific, so people tended to put the city they were in, though some just specified their country and some got as specific as their street address. Again, sometimes romanized and sometimes written in their native language. Like `language`, we stopped collecting this field when we transitioned from MITx to edX, so it's only available for our first batch of students.
location
----------
No longer used.
`meta`
**History**: User's location, asked during the sign up process for the 6.002x prototype course given in the Spring of 2012. The request was not specific, so people tended to put the city they were in, though some just supplied a country and some got as specific as their street address. Again, sometimes romanized and sometimes written in their native language. Like ``language``, edX stopped collecting this column after MITx transitioned to edX, so it is only available for the first batch of students.
------
meta
------
An optional, freeform text field that stores JSON data. This was a hack to allow us to associate arbitrary metadata with a user. An example of the JSON that can be stored here is::
{
"old_names" : [
["Mike Smith", "Mike's too informal for a certificate.", "2012-11-15T17:28:12.658126"],
["Michael Smith", "I want to add a middle name as well.", "2013-02-07T11:15:46.524331"]
"improvement_textbook": ["I'd like to get the full PDF."],
"future_offerings": ["true"],
"university_comparison":
["This course was <strong>on the same level</strong> as the university class."],
"improvement_lectures": ["More PowerPoint!"],
"highest_degree": ["Bachelor's degree."],
"future_classes": ["true"],
"future_updates": ["true"],
"favorite_parts": ["Releases, bug fixes, and askbot."]
}
}
The following are details about this metadata. Please note that the fields described below are found as JSON attributes *inside* the `meta` field, and are *not* separate database fields of their own.
`old_names`
A list of the previous names this user had, and the timestamps at which they submitted a request to change those names. These name change request submissions used to require a staff member to approve it before the name change took effect. This is no longer the case, though we still record their previous names.
An optional, freeform text field that stores JSON data. This field allows us to associate arbitrary metadata with a user. An example of the JSON that can be stored in this field follows, using pretty print for an easier-to-read display format.
.. code-block:: json
{
"old_names": [
[
"Mike Smith",
"Mike's too informal for a certificate.",
"2012-11-15T17:28:12.658126"
],
[
"Michael Smith",
"I want to add a middle name as well.",
"2013-02-07T11:15:46.524331"
]
],
"old_emails": [
[
"mr_mike@email.com",
"2012-10-18T15:21:41.916389"
]
],
"6002x_exit_response": {
"rating": [
"6"
],
"teach_ee": [
"I do not teach EE."
],
"improvement_textbook": [
"I'd like to get the full PDF."
],
"future_offerings": [
"true"
],
"university_comparison": [
"This course was <strong>on the same level<\/strong> as the university class."
],
"improvement_lectures": [
"More PowerPoint!"
],
"highest_degree": [
"Bachelor's degree."
],
"future_classes": [
"true"
],
"future_updates": [
"true"
],
"favorite_parts": [
"Releases, bug fixes, and askbot."
]
}
}
Details about this metadata follow. Please note that the "fields" described here are found as JSON attributes *inside* a given ``meta`` field, and are *not* separate database columns of their own.
``old_names``
A list of the previous names this user had, and the timestamps at which they submitted a request to change those names. These name change request submissions used to require a staff member to approve it before the name change took effect. This is no longer the case, though their previous names are still recorded.
Note that the value stored for each entry is the name they had, not the name they requested to get changed to. People often changed their names as the time for certificate generation approached, to replace nicknames with their actual names or correct spelling/punctuation errors.
The timestamps are UTC, like all datetimes stored in our system.
`old_emails`
``old_emails``
A list of previous emails this user had, with timestamps of when they changed them, in a format similar to `old_names`. There was never an approval process for this.
The timestamps are UTC, like all datetimes stored in our system.
`6002x_exit_response`
Answers to a survey that was sent to students after the prototype 6.002x course in the Spring of 2012. The questions and number of questions were randomly selected to measure how much survey length affected response rate. Only students from this course have this field.
``6002x_exit_response``
Answers to a survey that was sent to students after the prototype 6.002x course in the Spring of 2012. The questions and number of questions were randomly selected to measure how much survey length affected response rate. Only students from this course have this field.
`courseware`
------------
This can be ignored. At one point, it was part of a way to do A/B tests, but it has not been used for anything meaningful since the conclusion of the prototype course in the spring of 2012.
courseware
------------
No longer used.
`gender`
**History**: At one point, it was part of a way to do A/B tests, but it has not been used for anything meaningful since the conclusion of the prototype course in the spring of 2012.
--------
gender
--------
Dropdown field collected during student signup. We only started collecting this information after the transition from MITx to edX, so prototype course students will have `NULL` for this field.
Collected during student signup from a dropdown list control.
.. list-table::
:widths: 10 80
:header-rows: 1
:widths: 10 80
:header-rows: 1
* - Value
- Description
* - f
- Female
* - m
- Male
* - o
- Other
* - (blank)
- User did not specify a gender.
* - NULL
- This student signed up before this information was collected.
**History**: This information began to be collected after the transition from MITx to edX; prototype course students have NULL for this field.
* - Value
- Meaning
* - `NULL`
- This student signed up before this information was collected
* - `''` (blank)
- User did not specify gender
* - `'f'`
- Female
* - `'m'`
- Male
* - `'o'`
- Other
`mailing_address`
-----------------
Text field collected during student signup. We only started collecting this information after the transition from MITx to edX, so prototype course students will have `NULL` for this field. Students who elected not to enter anything will have a blank string.
mailing_address
-----------------
Collected during student signup from a text field control. A blank string for students who elect not to enter anything.
`year_of_birth`
**History**: This information began to be collected after the transition from MITx to edX; prototype course students have NULL for this field.
---------------
year_of_birth
---------------
Dropdown field collected during student signup. We only started collecting this information after the transition from MITx to edX, so prototype course students will have `NULL` for this field. Students who decided not to fill this in will also have NULL.
Collected during student signup from a dropdown list control. NULL for students who decide not to fill this in.
`level_of_education`
**History**: This information began to be collected after the transition from MITx to edX; prototype course students have NULL for this field.
--------------------
level_of_education
--------------------
Dropdown field collected during student signup. We only started collecting this information after the transition from MITx to edX, so prototype course students will have `NULL` for this field.
Collected during student signup from a dropdown list control.
.. list-table::
:widths: 10 80
:header-rows: 1
:widths: 10 80
:header-rows: 1
* - Value
- Description
* - p
- Doctorate.
* - m
- Master's or professional degree.
* - b
- Bachelor's degree.
* - a
- Associate's degree.
* - hs
- Secondary/high school.
* - jhs
- Junior secondary/junior high/middle school.
* - el
- Elementary/primary school.
* - none
- None.
* - other
- Other.
* - (blank)
- User did not specify level of education.
* - p_se
- Doctorate in science or engineering (no longer used).
* - p_oth
- Doctorate in another field (no longer used).
* - NULL
- This student signed up before this information was collected.
**History**: Data began to be collected in this column after the transition from MITx to edX; prototype course students have NULL for this field.
* - Value
- Meaning
* - `NULL`
- This student signed up before this information was collected
* - `''` (blank)
- User did not specify level of education.
* - `'p'`
- Doctorate
* - `'p_se'`
- Doctorate in science or engineering (no longer used)
* - `'p_oth'`
- Doctorate in another field (no longer used)
* - `'m'`
- Master's or professional degree
* - `'b'`
- Bachelor's degree
* - `'a'`
- Associate's degree
* - `'hs'`
- Secondary/high school
* - `'jhs'`
- Junior secondary/junior high/middle school
* - `'el'`
- Elementary/primary school
* - `'none'`
- None
* - `'other'`
- Other
`goals`
-------
Text field collected during student signup in response to the prompt, "Goals in signing up for edX". We only started collecting this information after the transition from MITx to edX, so prototype course students will have `NULL` for this field. Students who elected not to enter anything will have a blank string.
goals
-------
Collected during student signup from a text field control with the label "Goals in signing up for edX". A blank string for students who elect not to enter anything.
`allow_certificate`
**History**: This information began to be collected after the transition from MITx to edX; prototype course students have NULL for this field.
-------------------
allow_certificate
-------------------
Set to `1` for most students. This field is set to `0` if log analysis has revealed that this student is accessing our site from a country that the US has an embargo against. At this time, we do not issue certificates to students from those countries.
Set to 1 (true).
**History**: Prior to 10 Feb 2014, this field was set to 0 (false) if log analysis revealed that the student was accessing the edX site from a country that the U.S. had embargoed. This restriction is no longer in effect, and on 10 Feb 2014 this value was changed to 1 for all users.
.. _student_courseenrollment:
==============================================
Columns in the student_courseenrollment Table
==============================================
A row in this table represents a student's enrollment for a particular course run.
note:: A row is created for every student who starts the enrollment process, even if they never complete registration.
`student_courseenrollment`
==========================
A row in this table represents a student's enrollment for a particular course run. If they decide to unenroll in the course, we set `is_active` to `False`. We still leave all their state in `courseware_studentmodule` untouched, so they will not lose courseware state if they unenroll and reenroll.
**History**: As of 20 Aug 2013, this table retains the records of students who unenroll. Records are no longer deleted from this table.
`id`
A sample of the heading row and a data row in the ``student_courseenrollment`` table follow.
The ID of the course run they're enrolling in (e.g. `MITx/6.002x/2012_Fall`). You can get this from the URL when you're viewing courseware on your browser.
course_id
-----------
The ID of the course run that the user is enrolling in (for example, MITx/6.002x/2012_Fall). You can get this from the URL when you view the courseware on your browser.
`created`
---------
Datetime of enrollment, UTC.
created
---------
Stores the date and time that this row was created, in UTC format.
`is_active`
-----------
Boolean indicating whether this enrollment is active. If an enrollment is not active, a student is not enrolled in that course. This lets us unenroll students without losing a record of what courses they were enrolled in previously. This was introduced in the 2013-08-20 release. Before this release, unenrolling a student simply deleted the row in `student_courseenrollment`.
is_active
-----------
Boolean indicating whether this enrollment is active. If an enrollment is not active, a student is not enrolled in that course. For example, if a student decides to unenroll from the course, ``is_active`` is set to 0 (false). The student's state in ``courseware_studentmodule`` is untouched, so courseware state is not lost if a student unenrolls and then re-enrolls.
``is_active`` can also be set to 0 if a student begins the process of enrolling in a course by purchasing a verified certificate, but then abandond the shopping cart before completing the purchase (and the enrollment).
**History**: This column was introduced in the 20 Aug 2013 release. Before this release, unenrolling a student simply deleted the row in ``student_courseenrollment``.
`mode`
------
String indicating what kind of enrollment this was. The default is "honor" (honor certificate) and all enrollments prior to 2013-08-20 will be of that type. Other types being considered are "audit" and "verified_id".
mode
------
String indicating what kind of enrollment this is: blank, audit, honor, or verified.
**History**: All enrollments prior to 20 Aug 2013 are "honor".
.. _user_id_map:
==================================
Columns in the user_id_map Table
==================================
`user_id_map`
==========================
A row in this table maps a student's real user ID to an anonymous ID generated to obfuscate the student's identity.
A sample of the heading row and a data row in the ``user_id_map`` table follow.
.. code-block:: sql
.. list-table::
hash_id id username
e9989f2cca1d699d88e14fd43ccb5b5f 9999999 AAAAAAAA
The ``student_courseenrollment`` table has the following columns:
.. list-table::
:widths: 15 15 15 15
:header-rows: 1
* - Field
* - Column
- Type
- Null
- Key
...
...
@@ -400,283 +641,413 @@ A row in this table maps a student's real user ID to an anonymous ID generated t
- NO
-
`hash_id`
----
----------
hash_id
----------
The user ID generated to obfuscate the student's identity.
`user_id`
---------
The student's ID in `auth_user.id`.
id
---------
The student's ID in ``auth_user.id``.
`username`
-----------
The student's username in `auth_user.id`.
username
-----------
The student's username in ``auth_user.username``.
.. _Courseware_Progress:
************************
Courseware Progress Data
************************
Any piece of content in the courseware can store state and score in the ``courseware_studentmodule`` table. Grades and the user Progress page are generated by doing a walk of the course contents, searching for graded items, looking up a student's entries for those items in ``courseware_studentmodule`` via *(course_id, student_id, module_id)*, and then applying the grade weighting found in the course policy and grading policy files. Course policy files determine how much weight one problem has relative to another, and grading policy files determine how much categories of problems are weighted (for example, HW=50%, Final=25%, etc.).
*******************
Courseware Progress
*******************
==================================
About Modules
==================================
Any piece of content in the courseware can store state and score in the `courseware_studentmodule` table. Grades and the user Progress page are generated by doing a walk of the course contents, searching for graded items, looking up a student's entries for those items in `courseware_studentmodule` via `(course_id, student_id, module_id)`, and then applying the grade weighting found in the course policy and grading policy files. Course policy files determine how much weight one problem has relative to another, and grading policy files determine how much categories of problems are weighted (e.g. HW=50%, Final=25%, etc.).
It's important to understand what "modules" are in the context of our system, as the terminology can be confusing. For the conventions of this table and many parts of our code, a "module" is a content piece that appears in the courseware. This can be nearly anything that appears when users are in the courseware tab: a video, a piece of HTML, a problem, etc. Modules can also be collections of other modules, such as sequences, verticals (modules stacked together on the same page), weeks, chapters, etc. In fact, the course itself is a top level module that contains all the other contents of the course as children. You can imagine the entire course as a tree with modules at every node.
.. warning::
**Modules might not be what you expect!**
Modules can store state, but whether and how they do so varies based on the implementation for that particular kind of module. When a user loads a page, the system looks up all the modules that need to be rendered in order to display it, and then asks the database to look up state for those modules for that user. If there is no corresponding entry for that user for a given module, a new row is created and the state is set to an empty JSON dictionary.
It's important to understand what "modules" are in the context of our system, as the terminology can be confusing. For the conventions of this table and many parts of our code, a "module" is a content piece that appears in the courseware. This can be nearly anything that appears when users are in the courseware tab: a video, a piece of HTML, a problem, etc. Modules can also be collections of other modules, such as sequences, verticals (modules stacked together on the same page), weeks, chapters, etc. In fact, the course itself is a top level module that contains all the other contents of the course as children. You can imagine the entire course as a tree with modules at every node.
.. _courseware_studentmodule:
Modules can store state, but whether and how they do so is up to the implemenation for that particular kind of module. When a user loads page, we look up all the modules they need to render in order to display it, and then we ask the database to look up state for those modules for that user. If there is corresponding entry for that user for a given module, we create a new row and set the state to an empty JSON dictionary.
The ``courseware_studentmodule`` table holds all courseware state for a given user.
`courseware_studentmodule`
==========================
A sample of the heading row and a data row in the ``courseware_studentmodule`` table follow.
The `courseware_studentmodule` table holds all courseware state for a given user. Every student has a separate row for every piece of content in the course, making this by far our largest table::
.. code-block:: sql
+-------------+--------------+------+-----+
| Field | Type | Null | Key |
+-------------+--------------+------+-----+
| id | int(11) | NO | PRI |
| module_type | varchar(32) | NO | MUL |
| module_id | varchar(255) | NO | MUL |
| student_id | int(11) | NO | MUL |
| state | longtext | YES | |
| grade | double | YES | MUL | # problem, selfassessment, and combinedopenended use this
| created | datetime | NO | MUL |
| modified | datetime | NO | MUL |
| max_grade | double | YES | | # problem, selfassessment, and combinedopenended use this
| done | varchar(8) | NO | MUL | # ignore this
| course_id | varchar(255) | NO | MUL |
+-------------+--------------+------+-----+
id module_type module_id student_id state grade created modified max_grade done
2013-03-19 17:21:07 2014-01-07 20:18:54 NULL na edX/DemoX/Demo_course
`id`
Students have a separate row for every piece of content that they access or that is created to hold state data, making this the largest table in the data package.
The ``courseware_studentmodule`` table has the following columns:
Primary key. Rarely used though, since most lookups on this table are searches on the three tuple of `(course_id, student_id, module_id)`.
`module_type`
-------------
module_type
-------------
.. list-table::
:widths: 10 80
:header-rows: 0
:widths: 20 70
:header-rows: 1
* - `chapter`
* - Type
- Description
* - chapter
- The top level categories for a course. Each of these is usually labeled as a Week in the courseware, but this is just convention.
* - `combinedopenended`
- A new module type developed for grading open ended questions via self assessment, peer assessment, and machine learning.
* - `conditional`
- A new module type recently developed for 8.02x, this allows you to prevent access to certain parts of the courseware if other parts have not been completed first.
* - `course`
* - combinedopenended
- A module type developed for grading open ended questions via self assessment, peer assessment, and machine learning.
* - conditional
- Allows you to prevent access to certain parts of the courseware if other parts have not been completed first.
* - course
- The top level course module of which all course content is descended.
* - `problem`
- A problem that the user can submit solutions for. We have many different varieties.
* - `problemset`
- A collection of problems and supplementary materials, typically used for homeworks and rendered as a horizontal icon bar in the courseware. Use is inconsistent, and some courses use a `sequential` instead.
* - `selfassessment`
- Self assessment problems. An early test of the open ended grading system that is not in widespread use yet. Recently deprecated in favor of `combinedopenended`.
* - `sequential`
* - crowdsource_hinter
- Not currently used. **History**: This ``module_type`` was included in a single course on a test basis and then deprecated.
* - lti
- Learning Tools Interoperability component that adds an external learning application to display content, or to display content and also require a student response.
* - peergrading
- Indicates a problem that is graded by other students. An option for grading open ended questions.
* - poll_question
- Not currently used. **History**: This ``module_type`` was included in a single course on a test basis and then deprecated.
* - problem
- A problem that the user can submit solutions for. EdX offers many different varieties.
* - problemset
- A collection of problems and supplementary materials, typically used for homeworks and rendered as a horizontal icon bar in the courseware. Use is inconsistent, and some courses use a ``sequential`` instead.
* - randomize
- Identifies a module in which one of several possible defined alternatives is randomly selected for display to each student.
* - selfassessment
- Self assessment problems. Used in a single course in Fall 2012 as an early test of the open ended grading system. Deprecated in favor of ``combinedopenended``.
* - sequential
- A collection of videos, problems, and other materials, rendered as a horizontal icon bar in the courseware.
* - `videosequence`
- A collection of videos, exercise problems, and other materials, rendered as a horizontal icon bar in the courseware. Use is inconsistent, and some courses use a `sequential` instead.
* - timelimit
- Not currently used. **History**: This ``module_type`` was included in a single course on a test basis and then deprecated.
* - video
- A component that makes a video file available for students to play.
* - videoalpha
- Not currently used. **History**: During the implementation of a change to the ``video`` ``module_type``, both ``video`` and ``videoalpha`` were stored. The ``videoalpha`` type was then deprecated.
* - videosequence
- A collection of videos, exercise problems, and other materials, rendered as a horizontal icon bar in the courseware. **History**: This ``module_type`` is no longer in use, courses now use ``sequential`` instead.
* - word_cloud
- A specialized problem that produces a graphic from the words that students enter.
There's been substantial muddling of our container types, particularly between sequentials, problemsets, and videosequences. In the beginning we only had sequentials, and these ended up being used primarily for two purposes: creating a sequence of lecture videos and exercises for instruction, and creating homework problem sets. The `problemset` and `videosequence` types were created with the hope that our system would have a better semantic understanding of what a sequence actually represented, and could at a later point choose to render them differently to the user if it was appropriate. Due to a variety of reasons, migration over to this has been spotty. They all render the same way at the moment.
`module_id`
-----------
Unique ID for a distinct piece of content in a course, these are recorded as URLs of the form `i4x://{org}/{course_num}/{module_type}/{module_name}`. Having URLs of this form allows us to give content a canonical representation even as we are in a state of transition between backend data stores.
module_id
-----------
Unique ID for a distinct piece of content in a course, these are recorded as URLs of the form i4x://{org}/{course_num}/{module_type}/{module_name}. Having URLs of this form allows us to give content a canonical representation even during a transition between backend data stores.
.. list-table:: Breakdown of example `module_id`: `i4x://MITx/3.091x/problemset/Sample_Problems`
:widths: 10 20 70
As an example, this module_id:
i4x://MITx/3.091x/problemset/Sample_Problems
contains the following parts:
.. list-table::
:widths: 15 20 55
:header-rows: 1
* - Part
- Example
- Definition
* - `i4x://`
* - i4x://
-
- Just a convention we ran with. We had plans for the domain `i4x.org` at one point.
* - `org`
- `MITx`
- A convention selected based on plans for an i4x.org domain.
* - org
- MITx
- The organization part of the ID, indicating what organization created this piece of content.
* - `course_num`
- `3.091x`
- The course number this content was created for. Note that there is no run information here, so you can't know what runs of the course this content is being used for from the `module_id` alone; you have to look at the `courseware_studentmodule.course_id` field.
* - `module_type`
- `problemset`
- The module type, same value as what's in the `courseware_studentmodule.module_type` field.
* - `module_name`
- `Sample_Problems`
- The name given for this module by the content creators. If the module was not named, the system will generate a name based on the type and a hash of its contents (ex: `selfassessment_03c483062389`).
`student_id`
* - course_num
- 3.091x
- The course number this content was created for. Note that there is no run information here, so you can't know what runs of the course this content is being used for from the ``module_id`` alone; you have to look at the ``courseware_studentmodule.course_id`` column.
* - module_type
- problemset
- The module type, same value as found in the ``courseware_studentmodule.module_type`` column.
* - module_name
- Sample_Problems
- The name given for this module by the content creators. If the module was not named, the system generates a name based on the type and a hash of its contents (for example, ``selfassessment_03c483062389``).
------------
A reference to `auth_user.id`, this is the student that this module state row belongs to.
student_id
------------
A reference to ``auth_user.id``, this is the student that this module state row belongs to.
`state`
-------
state
-------
This is a JSON text field where different module types are free to store their state however they wish.
The state for all of these is a JSON dictionary indicating the user's last known position within this container. This is 1-indexed, not 0-indexed, mostly because it went out that way at one point and we didn't want to later break saved navigation state for users.
When this user last interacted with this course/chapter/etc., they had clicked on the third child element. Note that the position is a simple index and not a `module_id`, so if you rearranged the order of the contents, it would not be smart enough to accomodate the changes and would point users to the wrong place.
The state for all of these container modules is a JSON dictionary indicating the user's last known position within this container. This is 1-indexed, not 0-indexed, mostly because it was released that way and a later change would have broken saved navigation state for users.
JSON data that shows the status of Open Response Assessment problems.
When this user last interacted with this course/chapter/etc., they clicked on the third child element. Note that the position is a simple index and not a ``module_id``, so if you rearranged the order of the contents, it would not be smart enough to accomodate the changes and would point users to the wrong place.
The problem can be in one of four states:
* **initial**
* **assessing**
* **post_assessment**
* **done**
The hierarchy of these containers is ``course > chapter > (problemset | sequential | videosequence)``
Following is an example of a problem in **done** state:
``combinedopenended``
::
The JSON document includes attributes that identify the student's ``answer``, a ``rubric_xml`` that includes the complete XML syntax for the rubric, the ``score`` earned and the ``max_score``, and the ``grader_id`` (the ``auth_user.id``) of each student who assessed the answer.
There are many kinds of problems supported by the system, and they all have different state requirements. Note that a single problem can have many different response fields. If a problem generates a random circuit and asks five questions about it, then all of that is stored in one row in ``courseware_studentmodule``.
`conditional`
Conditionals don't actually store any state, so this value is always an empty JSON dictionary (`'{}'`). We should probably remove these entries altogether.
.. Include the different problem types and info about the state.
`problem`
There are many kinds of problems supported by the system, and they all have different state requirements. Note that one problem can have many different response fields. If a problem generates a random circuit and asks five questions about it, then all of that is stored in one row in `courseware_studentmodule`.
``selfassessment``
TODO: Write out different problem types and their state.
In the course that used this module type, the JSON document included attributes for the ``student_answers``, the ``scores`` earned and ``max_score``, and any ``hints`` provided.
`selfassessment`
TODO: More details to come.
`grade`
-------
grade
-------
Floating point value indicating the total unweighted grade for this problem that the student has scored. Basically how many responses they got right within the problem.
Only `problem` and `selfassessment` types use this field. All other modules set this to `NULL`. Due to a quirk in how rendering is done, `grade` can also be `NULL` for a tenth of a second or so the first time that a user loads a problem. The initial load will trigger two writes, the first of which will set the `grade` to `NULL`, and the second of which will set it to `0`.
Only ``problem`` and ``selfassessment`` types use this column. All other modules set this to NULL. Due to a quirk in how rendering is done, ``grade`` can also be NULL for a tenth of a second or so the first time that a user loads a problem. The initial load triggers two writes, the first of which sets the ``grade`` to NULL, and the second of which sets it to 0.
`created`
---------
Datetime when this row was created (i.e. when the student first accessed this piece of content).
created
---------
Datetime when this row was created, which is typically when the student first accesses this piece of content.
**Note**: For a module that contains multiple child modules, a row is created for each of them when the student first accesses one of them.
`modified`
----------
Datetime when we last updated this row. Set to be equal to `created` at first. A change in `modified` implies that there was a state change, usually in response to a user action like saving or submitting a problem, or clicking on a navigational element that records its state. However it can also be triggered if the module writes multiple times on its first load, like problems do (see note in `grade`).
modified
----------
Datetime when this row was last updated. Set to be equal to ``created`` at first. A change in ``modified`` implies that there was a state change, usually in response to a user action like saving or submitting a problem, or clicking on a navigational element that records its state. However it can also be triggered if the module writes multiple times on its first load, like problems do (see note in ``grade``).
`max_grade`
-----------
Floating point value indicating the total possible unweighted grade for this problem, or basically the number of responses that are in this problem. Though in practice it's the same for every entry with the same `module_id`, it is technically possible for it to be anything. The problems are dynamic enough where you could create a random number of responses if you wanted. This a bad idea and will probably cause grading errors, but it is possible.
Another way in which `max_grade` can differ between entries with the same `module_id` is if the problem was modified after the `max_grade` was written and the user never went back to the problem after it was updated. This might happen if a member of the course staff puts out a problem with five parts, realizes that the last part doesn't make sense, and decides to remove it. People who saw and answered it when it had five parts and never came back to it after the changes had been made will have a `max_grade` of `5`, while people who saw it later will have a `max_grade` of `4`.
max_grade
-----------
Floating point value indicating the total possible unweighted grade for this problem, or basically the number of responses that are in this problem. Though in practice it's the same for every entry with the same ``module_id``, it is technically possible for it to be anything.
These complexities in our grading system are a high priority target for refactoring in the near future.
Another way in which ``max_grade`` can differ between entries with the same ``module_id`` is if the problem was modified after the ``max_grade`` was written and the user never went back to the problem after it was updated. This might happen if a member of the course staff puts out a problem with five parts, realizes that the last part doesn't make sense, and decides to remove it. People who saw and answered it when it had five parts and never came back to it after the changes had been made will have a ``max_grade`` of 5, while people who saw it later will have a ``max_grade`` of 4.
Only `problem` and `selfassessment` types use this field. All other modules set this to `NULL`.
Only graded module types use this column, with ``problem`` being the primary example. All other modules set this to NULL.
`done`
------
Ignore this field. It was supposed to be an indication whether something was finished, but was never properly used and is just `'na'` in every row.
done
------
Not used. The value ``na`` appears in every row.
`course_id`
-----------
The course that this row applies to, represented in the form org/course/run (ex: `MITx/6.002x/2012_Fall`). The same course content (same `module_id`) can be used in different courses, and a student's state needs to be tracked separately for each course.
course_id
-----------
The course that this row applies to, represented in the format org/course/run (for example, ``MITx/6.002x/2012_Fall``). The same course content (same ``module_id``) can be used in different courses, and a student's state needs to be tracked separately for each course.
The ``certificates_generatedcertificate`` table tracks the state of certificates and final grades for a course. The table is populated when a script is run to grade all of the students who are enrolled in the course at that time and issue certificates. The certificate process can be rerun and this table is updated appropriately.
A sample of the heading row and two data rows in the ``certificates_generatedcertificate`` table follow.
.. code-block:: sql
id user_id download_url grade course_id key distinction status verify_uuid
download_uuid name created_date modified_date error_reason mode
The ``download_url`` contains the full URL to the certificate.
The generatedcertificate table tracks certificate state for students who have been graded after a course completes. Currently the table is only populated when a course ends and a script is run to grade students who have completed the course::
-------
grade
-------
The grade computed the last time certificate generation ran. If the courseware, student state, or grading policy change, the value in this column can be different than the grade shown on a student's Progress page.
Used internally only. A random string that is used to match server requests to responses sent to the LMS.
-----------------
distinction
-----------------
Not used.
`user_id`, `course_id`
----------------------
The table is indexed by user and course
**History**: This was used for letters of distinction for 188.1x, but is not being used for any current courses.
`status`
--------
Status may be one of these states:
* `unavailable`
* `generating`
* `regenerating`
* `deleting`
* `deleted`
* `downloadable`
* `notpassing`
* `restricted`
* `error`
After a course has been graded and certificates have been issued status will be one of:
status
--------
Status can be one of these states:
* `downloadable`
* `notpassing`
* `restricted`
.. list-table::
:widths: 15 80
:header-rows: 1
* - Value
- Description
* - deleted
- The certificate has been deleted.
* - deleting
- A request has been made to delete a certificate.
* - downloadable
- The student passed the course and a certificate is available for download.
* - error
- An error ocurred during certificate generation.
* - generating
- A request has been made to generate a certificate but it has not yet been generated.
* - notpassing
- The student's grade is not a passing grade.
* - regenerating
- A request has been made to regenerate a certificate but it has not yet been generated.
* - restricted
- No longer used. **History**: Specified when ``userprofile.allow_certificate`` was set to false: to indicate that the student was on the restricted embargo list.
* - unavailable
- No entry, typically because the student has not yet been graded for certificate generation.
After a course has been graded and certificates have been issued, status is one of:
* downloadable
* notpassing
If the status is `downloadable` then the student passed the course and there will be a certificate available for download.
-------------
verify_uuid
-------------
A hash code that verifies the validity of a certificate. Included on the certificate itself as part of a URL.
-------------
download_uuid
-------------
A hash code that identifies this student's certificate. Included as part of the ``download_url``.
`download_url`
--------------
The `download_uuid` has the full URL to the certificate
------
name
------
This column records the name of the student that was set at the time the student was graded and the certificate was generated.
`download_uuid`, `verify_uuid`
------------------------------
The two uuids are what uniquely identify the download url and the url used to download the certificate.
---------------
created_date
---------------
Date this row in the database was created.
`distinction`
-------------
This was used for letters of distinction for 188.1x and is not being used for any current courses
---------------
modified_date
---------------
Date this row in the database was modified.
`name`
------
This field records the name of the student that was set at the time the student was graded and the certificate was generated.
---------------
error_reason
---------------
Used internally only. Logs messages that are used for debugging if the certificate generation process fails.
`grade`
-------
The grade of the student recorded at the time the certificate was generated. This may be different than the current grade since grading is only done once for a course when it ends.
---------------
mode
---------------
Contains the value found in the ``enrollment.mode`` field for a student and course at the time the certificate was generated: blank, audit, honor, or verified. This value is not updated if the student's ``enrollment.mode`` changes after certificates are generated.
