Skip to content

Measures Catalogue

Jump to bottom
alanepaull edited this page Aug 5, 2019 · 20 revisions

Table of Contents

List of Measures

1 Unstructured feedback

Unstructured text (or speech-to-text) captured from a range of methods.

Methods Online form on kiosk
Online form on mobile
Online form on VLE or website
Microphone (connected to voice recognition software)
Passive social media harvesting
Contexts Time
Location (if method has geo metadata)
Place (if locations are mapped)
Module (if time and place connected to timetable)
Student (if student ID is used to activate recording or is passively captured)
Student on Module Instance (if student and module both captured)
Passive social media harvesting can make use of context data captured at source, as well as hashtags and QR codes.
Algorithms Unstructured feedback analysis (e.g. Keatext https://www.keatext.ai/)
Sentiment analysis (e.g. Lexalytics: https://www.lexalytics.com/technology/sentiment)
Text mining
Privacy Form-based capture requires explicit permission.
Mic-based capture has serious privacy issues.
Passive social media harvesting has serious privacy issues.
Activity sources VLE
CRM
Kiosk software
Mobile feedback apps
Social media sites

2 Structured feedback

This can be either a simple “happiness rating”, or a rating in response to a specific contextual prompt, e.g. “How safe does it feel on campus?”, “How are the lunch queues today?”

Methods Dedicated physical button device
Online form on kiosk
Online form on mobile
Online form on VLE or website
Contexts Time
Location (if method has geo metadata)
Place (if locations are mapped)
Module (if time and place connected to timetable)
Student (if student ID is used to activate)
Student on Module Instance (if student and module are both captured)
Algorithms This measure lends itself to all standard forms of machine learning and analytics, such as linear regression, logistic regression, etc.
Privacy Form-based capture requires explicit permission.
Possible issues if identity is passively captured rather than user-initiated
Activity sources HappyOrNot terminals: https://www.happy-or-not.com/en/
VLE
CRM
Kiosk software

Notes: Physical button feedback devices are typically anonymous and placed in general use locations, so only provide limited context. However, it is possible they could read a student or staff ID card by RFID or NFC to obtain an identity. They could also be placed in existing teaching spaces rather than general use areas.

3 Sentiment/emotion and facial recognition

Passive capture of emotion or sentiment. Can be derived from a range of modalities - textual, auditory, and visual.

Methods Passive microphone
Passive social media
Passive communications monitoring
Passive facial / emotional recognition
CCTV and such like (doesn’t have to be passive)
Online form on kiosk
Online form on mobile
Online form on VLE or website
Scanning of students' faces, particularly in libraries and other hot-spots
Contexts Passive Camera
Time
Location (if method has geo metadata)
Place (if locations are mapped)
Module (if time and place connected to timetable)
Student (if student ID is used to activate or is passively captured)
Student on Module Instance (if student and module are both captured)
Requires initial image of face in system.
Algorithms Textual sentiment analysis, e.g. Lexalytics: https://www.lexalytics.com/technology/sentiment
Facial emotion recognition, e.g. NViso: http://www.nviso-insights.com/en
Audio emotion detection, e.g. Affectiva https://www.affectiva.com
Match of captured image in database of images. Link to attendance system.
Privacy Potentially serious concerns if capture is not triggered by the subject but is collected passively and context is also collected and retained.
Passive facial / emotional recognition has extreme privacy issues.
Activity sources VLE
Email
Kiosk software
Security camera system
Social media sites

4 Formal assessments

Grades and results that have been given or award from an institution. Not limited to results from the current institution. May also include timetable information.

Methods Grading systems
Entry information
Contexts Student
Subject
Time
Place of study
Grade
Student on Module Instance (if student and module are both captured)
Algorithms
Privacy Concerns regarding results and grades of student
Activity sources VLE Timetabling system
Student Record System

5 Environment monitoring

Environment monitoring collects data about the environment on campus.

Methods IoT connected hardware to collect environmental data
CCTV
Thermometers
Hygrometers
Microphone
CO2 monitoring device

Third party environmental data collection such as Met Office data points
Contexts Time
Location
Occupancy
Temperature
Humidity
CO2 levels
Lighting levels
Sound and noise levels
Network and wireless usage
Algorithms
Privacy Data should be about environment regarding buildings and places and should not be related to an individual person.
Activity sources

6 Traffic monitoring

Methods >Unspecified in the Use Case, presumably traffic data collection via wifi and traffic monitoring, coupled with analysis of flows over time. Wifi “presence” data.
Road traffic via cameras (video monitoring and analysis)
Various types of road sensors (pneumatic, piezoelectric, inductive loop, magnetic, acoustic, passive infrared, microwave and radar)
Contexts Student and staff location data Integration: access to / from city data hub
Algorithms Links to traffic lights. Congestion notifications to users.
Privacy Likely that anonymised data only is needed, so less privacy issue. “Issues of privacy and security will be of great concern to all of the stakeholders.”
Activity sources City data hub
Other road traffic nodes in campus or outside campus
Wifi nodes

7 Assessment of needs of students with disabilities

Methods Personal assessment and categorisation. Suggests an ontology of “disabilities” - academic work already done on this, particularly in relation to elearning. Linked to relevant ontology of learning styles that may be specific to disabilities, for example text-to-speech technologies.
Contexts Disability policies. Individual needs assessments usually required. Increasing number of students with disabilities, and recognition of need to handle demand. Pressure on resources, so intelligent solutions seen as needed. Greater emphasis on equality of access.
Algorithms
Privacy Paramount. Some individuals do not want to participate. “Students need to be confident that their information is kept private and shared appropriately with service providers. For universities there may be challenges in deciding on priorities and technical implementation of a wide range of devices and adjustments.”
Activity sources HEP systems containing disability data. Student_course_membership. Main source would be university assessments of needs at induction. Depends on student co-operation. Likely to be qualitative, rather than quantitative. Use case is very general, but there are specific technologies in use already - for example, text-to-speech and speech-to-text. These suggest analysis of use of the systems over time, so as to prioritise the most important.

8 Assessment of students' circumstances

Personal assessment and categorisation of social and personal circumstances

Methods Banking/financial data
Student records system
Contexts Student
Type of circumstance (health, financial)
Weighting
Algorithms
Privacy Paramount. Some individuals do not want to participate. Contains data about the student outside the institution
Activity sources Personal banking applications
Student records system
Health apps and fitness trackers

9 Physical well-being

Methods Health apps
Fitness trackers
Student record system
Contexts Student
Health Attribute
Result
Algorithms Targeted Offers: use health combined with preferences and other data to create “healthy nudges” such as discount codes for healthy foods and drinks.
Privacy
Activity sources

10 Student preferences

Methods Online form on kiosk
Online form on mobile
Online form on VLE or website
Contexts Student
Preference type
Preferencing setting
Algorithms
Privacy Privacy of the student is a concern. Data will contain information on the services the students use as well as the preferences themselves
Activity sources

11 Environmental accessibility of locations

Methods Heat and light sensors with automatic adjustment triggered by automatically sensed student need based on individual. Visual and audio signal reactions relating to position (for example, at lifts).
Contexts Primarily geared to student location at appropriate level of granularity.
Algorithms
Privacy Paramount. Some individuals do not want to participate. “Students need to be confident that their information is kept private and shared appropriately with service providers. For universities there may be challenges in deciding on priorities and technical implementation of a wide range of devices and adjustments.” Environmental reaction for 1 individual may point up that individual’s disability.
Activity sources Sensors at locations. Students’ devices. With base-line and in real-time.

12 Student location

Methods Built-in mobile device functions
GPS, mapping, inertial systems
Mobile phone transmitter tower locations
Wi-fi certified location and other wi-fi positioning systems
Devices' internal sensors - inertia, compass, accelerometer and gyroscope
Bluetooth beacons
Physical addresses and postcodes
Contexts Location: geo-location broadly
Place (within a geo-location): Accuracy of pin-pointing > depends on technology used; integration with maps (if any)
Module or course: Learning opportunity (module), or ad hoc learning - informal learning link (not sure how this context could be captured)?
Student
Student on module instance
Student on course instance
Institution / inst_tier: campus-based location data
City (or other broader location context than campus / institution)
Algorithms
Privacy Data capture for explicit tracking of individuals
Anonymisation
How is location data to be used in relation to interventions or other actions?
Activity sources Mobile devices
Wi-fi nodes and other campus infrastructure sensors
Bluetooth nodes
Traffic flow data
More devices, including wearable devices and implants. These are likely to be similar to smart phone sources?! Headsets, glasses, watches and wristbands.

13 Location

Methods QR codes (perhaps usage records location of student as the measure)?
Geo-location via GPS; integration with maps
Location, traffic and people flow data (eg open traffic data)
Augmented reality technologies
Mobile and wearable technologies
Simultaneous localization and mapping (SLAM)
Contexts Location: geo-location broadly
Place (within a geo-location): Accuracy of pin-pointing > depends on technology used; integration with maps (if any)
Module or Course: Learning opportunity (module), or ad hoc learning - informal learning link (not sure how this context could be captured)?
Student
Student on module instance
Student on course instance
Institution / inst_tier: campus-based location data
City (or other broader location context than campus / institution)
Algorithms Way-finding
Integration with maps
Integration with timetables
Privacy Data capture for explicit tracking of individuals
Anonymisation
How is location data to be used in relation to interventions or other actions?
Activity sources Mobile devices
Wi-fi nodes and other campus infrastructure sensors
Bluetooth nodes
Traffic flow data
More devices, including wearable devices and implants. These are likely to be similar to smart phone sources?! Headsets, glasses, watches and wristbands.

14 Occupancy

Using a range of methods, measure the level of occupancy of a space - this could be a room, but also a thoroughfare. This can be used for a range of purposes - for example locating less busy spaces for wayfinding, as well as adapting the heating and lighting based on the current occupancy levels.

Methods Passive infrared
Wi-fi triangulation
Swipe cards
Smart cards/lanyards and NFC
Contexts Time
Location (if method has geo metadata)
Place (if locations are mapped)
Module (if time and place connected to timetable)
Student (if student ID is passively captured, e.g. by NFC and smart card)
Student on Module Instance (if student and module both captured)
Bookings (if resource is bookable)
Algorithms Way-finding
Integration with maps
Integration with timetables
Predictive occupancy levels for timetabling and room booking or for environmental control, or for providing “busy times”
Information for users
Privacy
Activity sources Building management systems

15 Waste

Measure the volume, weight and/or composition of waste to help manage collection and disposal.

Methods Bins with built-in weighing scales or fill mark sensors
Using cameras to detect the level of ambient litter
QR codes to scan bins when they need emptying
QR codes on articles to aid choice of disposal
Contexts Time
Location (if method has geo metadata)
Place (if locations are mapped)
Algorithms Predicting time-to-full based on previous patterns to schedule preventative clean-up
Privacy Cameras may unintentionally capture individuals, so footage should not be retained.
Activity sources Waste management systems such as Smartbin: https://www.smartbin.com/markets/level-sensor-general-waste-recyclables/

16 Wear and usage

Measure the level of wear and/or usage of a facility to schedule preventative maintenance.

Methods Usage counters
Usage timers, for example on PCs
Built-in internal diagnostic sensors
Contexts Time
Location (if method has geo metadata)
Place (if locations are mapped)
Facility/Item
Algorithms Predictive preventative maintenance
Privacy Data should only relate to the facility, not the user
Activity sources Estate management systems
IT Service management systems

Many complex facilities already have built-in systems for this, such as vending machines and printers, that work with external services contracts. However, the same approach can be adapted for in-house facilities and systems without internal diagnostics.

17 Energy use

Using smart metering or similar, provide a measure of energy usage.

Methods Smart meters
Contexts Time
Location (depending on granularity of metering system)
Research infrastructure (again, depending on granularity)
Algorithms Predictive analytics for projected energy use.
Predictive pricing analytics for reducing energy cost.
Predictive analytics for yield from on-campus solar and wind generators (connecting with external weather data)
Privacy Not applicable
Activity sources Estate management systems.

18 Utilisation

The level of utilisation of a bookable or shared resource, such as parking, or PCs. This could be at a granular level (e.g. which PCs have been booked), aggregate level (e.g. % of parking spaces occupied), or on a waiting time basis (e.g. how long until a pod will become free).

Methods Direct reporting from information systems, such as timetabling and resource booking.
Contexts Time (both the start time and end time of booking are possible)
Location of the resource
Place (if locations are mapped)
Purpose of utilisation (if provided; for example, a room may be booked for a particular use, and this could update signage)
The type of resource
Module (if booking connected to module)
Student (if student ID is used in the booking)
Student on Module Instance (if student and module both captured)
Algorithms Predictive usage - for example how long it is likely to be until the resource is freed up.
Privacy Not applicable
Activity sources Timetabling
Other resource booking systems
Estate management

2.0 Blank Template

Methods
Contexts
Algorithms
Privacy
Activity sources

Clone this wiki locally