Reflections on #clouduc8 – Curation

Each time I am able to participate in Twitter Chat discussion I try to post a summary of the discussion to my blog. I do this both for my personal development as well as sharing with the Learning and Development Profession at large.

This summary is based on my own interpretations of the chat; others who participated may have differing opinions or interpretations of the discussion. I welcome those that do to add your ideas to the comments.

The topic of this month’s #clouduc8 session was Curation, and I was pleased to be the guest invited to submit the questions that guided the chat.

I find looking at the questions that are used to loosely guide the chat as a nice way to see the overall theme of the chat. Here are the discussion questions that were presented to the group:

1. What differences (if any) do you see between traditional curation and digital curation?
2. What tools are currently used for curation??
3. Can machines curate without human interaction? Why or why not? Where is the line drawn?
4. What additional features do machine-based tools need in order to better automate curation?

Curation is becoming one of the more common buzz words in the learning and performance field. This month’s Clouduc8 chat looked specifically at the tools and technology used for curation.

What are the differences between traditional curation and digital curation?

Curation in a digital environment is very different than more traditional curation environments such as museum or literary curation. For starters, the sheer volume of information available in digital form is staggering and increasing at an exponential rate. The deluge of information that we are presented with on a daily basis shifts the very foundation of curation from a place where it is a nice-to-have to an outright necessity to break through the noise and find what you are looking for.

In addition, digital curation requires a higher level of vetting than traditional curation. There’s a certain level of verification and fact-checking that is assumed to have taken place before a book or article is published in print. However, in the digital age we live in, anyone can publish digital content instantly, with little or no requirement that the information be accurate. Just because it’s online, doesn’t mean it’s true. Digital curation needs to account for that.

On the flip side, data in digital form enables use to leverage technology as a tool for curation. These tools, when used correctly, can automate some of the tasks associated with curation that previously were manual, and extremely time-consuming. Most data sources have search options that allow you to set criteria through which the aggregation and filtering of data can be automated. That eliminates a great deal of time that is generally spent on curation despite it providing little actual value. This allows the curator to spend their time on the tasks that add the true value to their curating.

Possibly the biggest difference between traditional and digital curation is defining who ‘qualifies’ as a curator. Traditional curating is a highly specialized skill. Museum curators are highly educated and trained in the skills of their craft. They are, indeed, professionals.

Digital technology – especially social media – has made it so anyone with an opinion or a desire to share an idea or story can easily be a curator. The age of the amateur curator is upon us; in a world where you can collect and publish ideas as easily as clicking a ‘Share’ button on facebook, anyone can be a curator.

What Tools are Currently Used for Curation?

There are a great number of tools that can be used for curation. In truth, any application that connects to a data source and has the ability to search and filter the data can be a tool for curation. One of the primary tools I use in my curation efforts is Twitter, as it is often the primary data source I am mining to curate resource pages.

There are also a growing number of tools that identify themselves as curation tools. As I mentioned earlier, curation is growing as a buzz word, and when that happens, you can be sure that tools that tap into that buzz will follow. Unfortunately, many of those tools can also miss the mark.

Two of the most popular tools labelled as curation tools are paper.li and scoop.it. These are excellent tools, but I would hesitate to identify them as curation tools. At best they aggregate and filter, which you could argue are the most basic forms of curation. However, I hesitate to label these as curation tools because they lack a critical component to the curation workflow: Human vetting.

Paper.li and Scoop.it enable you to set basic criteria (most often by linking to a Twitter list) and then services continually searches databases and pull out items that match the criteria. Worse yet, they also auto-publish.

Every day I get about 15-20 emails and tweets notifying me that someone’s paper.li or scoop.it page has been published. The individual doesn’t check the resources that have been pulled based on the criteria and manually vet them for content and to verify that the resources match the message or story the curator is looking to tell.

That’s what services like paper.li and scoop.it lack – the intervention by the curator to ensure the resources being shared match the purpose of sharing the resources in the first place… which is a requirement of true curation. There needs to be a purpose to it.

Compare those services with a service like Storify. This is a tool that truly enables curation, specifically curation for the purpose of telling a somewhat linear story. The tool pulls from a number of data sources including facebook, Twitter, Instagram, Youtube, and more, and allows the curator to pull data from each and organizae it in an order that tells a story. It has search options that can aggregate and filter the data to save time, but it requires the curator to manually intervene and add their value to the story.

Can machines curate without human interaction? Why or why not? Where is the line drawn?

This is a common question surrounding curation. You can only answer the question based on the technology of today, and today, the answer would be no. However, we should also be aware that the gap between what only humans can do and what machines can automate continues to shrink every day. I’ll take on any debate and argue that only humans can curate in today’s environment, but I will also say that it’s possible that could change at some point in the future.

But again, we don’t live in the future; we live in today. So where is the line drawn between human and machine capabilities related to curation?

Machines use in curation is really no different than the way we use machines in any other aspect of life. They are excellent tools to automate a task or process humans currently perform that requires little or no creative or problem solving thought processes.

When it comes to curation, machines can do a great deal – if not all – of the heavy lifting for aggregating and filtering of content. They struggle to do more than that though.

Consider Watson, the IBM computer that handily beat two champions in Jeopardy a few years ago. It was an amazing technical feat, and yet it also showed some of the gaps that exist between what machines can do in trying to understand and replicate human thought processes.

Jeopardy questions are unique. First, they are answers from which you need to determine the question. In addition, they often utilize puns and humor. In short, the meaning of each answer is rarely a literal translation of Language used to build it. There were a number of questions that Watson missed because it did not understand the subtle nuances of the human language.

What additional features do machine-based tools need in order to better automate curation?

Machines need to better understand the context of human language. Let’s say I wanted to curate a resource about the training field. A simple search and filtering on the word training would be effective.

It would not be enough though. It’s likely that such filtering would also include data on weight training, dog training, and other instances of the word training that have nothing to do with workplace learning and performance. Analytical software is doing an increasingly better job of recognizing this context, but it still has a way to go.

Machines understand structure very well, and can replicate tasks built on a structure with little or no error. But they do not understand content or context. Curation often involves being able to recognize relationships between two seemingly unrelated things. As machines begin to recognize potential relationships, they need to bubble them up to a human decision, and then learn from the answer.

Machines and technology are narrowing the gap between what they can automate and what requires human intervention, but I do not think humans are in any eminent danger of being removed from the curation equation.

First, curation automation is only as good as the algorithm it follows. That algorithm needs to be set by a human. From that perspective, even basic curation tasks like aggregation and filtering still require human intervention.

Second, and more importantly, the human element added to curation is often not about a structure or an algorithm, and therefore very difficult to automate. Even if machines could automate 95% of curation tasks, the remaining 5% would be a much larger part of overall output.

, , , , , ,