Digital Asset Management and AI for Sustainability Monitor
Kent academics explored a range of options for a data asset management system that could utilise machine learning to simplify and automate some business processes.
Sustainability Monitor (SusMon) works in the business intelligence industry and provides regular reports about sustainability to sell to companies who want to understand the breadth of sustainability strategies in their sector, in order to position their activity in relation to this. The objective of this project was to utilise opportunities presented by technology, and use them to organise data for forecasting purposes by considering how AI could be used to improve the company’s asset management and work flows.
This project was funded through the EIRA Innovation Voucher scheme and had two main aims:
- To look at the opportunities presented by technology with an emphasis on Digital Asset Management (DAM) systems for media and to improve the organisation of data in the company, supporting more efficient workflows.
To this end, the academic experts from the University of Kent’s School of Computing gave an overview of the kinds of features that are typically found in DAM systems, and identified which of these are important to the company in organising data better and improving workflows. For many companies with this kind of requirement there is a need to maintain a wide range of information: photographs, graphics, video, numerical data, spreadsheets, etc., so there are specific needs that require consideration. These findings were used to make a recommendation for next steps, and formulate a plan of immediate and future work.
- To look further into the future by considering how AI and machine learning could be used to improve the company’s asset management and workflows; in particular, information extraction work. This is work that is not routine, but which requires only a small amount of human supervision.
The project aims were for:
- The company to have a working DAM system which all staff can use
- Media assets to be easily shared across the company, rather than stored on individual machines
- Some repetitive tasks being automated
- A basic system for automating report creation to be trialled
- A plan for next steps to be created.
As one of the challenges is looking into asset management, a major part of this project will be implementing some kind of asset database for the company. One of the initial stages of the project looked at a number of features that are commonly found in DAM systems and related software, and considered the relevance of that to the SusMon workflow. These features included Transcoding, Brand Consistency, Version Control and Searchability.
Other potentially useful approaches to asset management included:
- Using local storage on worker’s computers – this is what is currently being done. Each person has the assets on their own machine, with sharing happening in an ad-hoc way e.g. through email.
- Using a shared file system, such as Google Drive. People can more readily share their documents, but there isn’t much detailed document / asset management provided.
- Using a generic document management system, such as SharePoint. This allows more definition of who is allowed to access documents and strong abilities to define workflows compared with a simple shared file system. However, such systems don’t contain specific facilities e.g. media processing.
- Commercial DAM systems for media management – there are lots of small providers in this space. Some are clearly targeted at large media organisations (e.g. TV / film production companies), some at e-commerce, some at smaller media / publishing operations. The most common model here is an annual subscription one. Some are very project / folder based; others have a model of being much more based around a large collection of searchable assets with heavy metadata tagging.
- Using an open-source, media-specific DAM – this is similar to the previous category, but potentially offers an advantage in that it could be modified to make it more specific to a particular need. The producers of these systems are typically commercial companies working on the principle of making the software available for free, but charging for hosting the system in the Cloud (though you can install it on your machine) and for support / training.
This project included an intern named Stephanie Daniels, from the School of Engineering & Arts. The SusMon website launched shortly before the start of the internship, which was also co-funded by EIRA. In turn, Stephanie has been responsible for making further tweaks to the system. This project also had a scheduled plan of work for Stephanie to complete.
All parties agreed that the project definitely delivered what the business laid out in their aims, and that Stephanie, in her role as project intern, proved invaluable to the successful implementation of the plan. This was a consultation project and was successful in providing the business with the information, guidance and insight to make their decision.
The internship has also had a significant impact on the business, as Ronnie McBryde, SusMon CEO explains:
“Data visualisation is key to Sustainability Monitor and fundamentally we are researchers, not designers. Having Stephanie in-house has enabled us to merge our research with design, which has allowed the company to move forward…Stephanie has worked on a wide range of projects, which have been possible because of her diverse set of skills.” – Ronnie McBryde, SusMon CEO
Reflecting on her internship, Stephanie commented that she ‘most enjoyed the designing element, being able to utilise the company’s existing branding and colour palette to create different charts and graphics. It’s great to work in a business environment and being able to apply my academic knowledge and skills’.
There are a number of future opportunities that Sustainability Monitor can explore, largely based around using AI and machine learning in their business processes. These build upon the above work, because once the company data is stored in a common place and, with good use of tagging and metadata, there will be both a source of data for these methods to work on, and a place to store the results discovered by an AI system so that they can be (manually or perhaps automatically) included in reports. Some initial suggestions were:
- Automating the extraction of numerical data from reports
- Automating the production of multiple report formats
- Automated extraction of information from social media
- Trend Analysis for social media
- Automated MetaData Targeting