One of the key changes recently implemented at Shaping Tomorrow is our new software that will not just add the basic details of an article but interpret the article with automatic suggestions for its implications. Walter Kehl developed this new tool for us, and we’ve asked him to write a little bit about what it is and how it can help Shaping Tomorrow’s users. This is the first installment, a basic overview. He has more to come so check back soon, and please feel free to ask any questions you might have in the comments.
When we look at what people do when they think about the future, especially when using a site like Shaping Tomorrow, we see that the main activity is the collection of knowledge about the future. More specifically, we see the current knowledge which could have an influence on our future: demographic and economic developments, market trends, scientific breakthroughs, technological innovation, new business ideas, lots and lots of information – “insights” – relevant in one way or another for our future. All these insights together can be seen as one big knowledge base about the future.
The way Shaping Tomorrow works is that users from all walks of life can contribute to this database according to their special interests. This has the big advantage that input is as varied and different as the user’s interests and purposes, but it brings with it the challenges that the resulting collection of pieces of knowledge is big, only loosely structured, and the individual insights are to a large degree unconnected. Also finding relevant information is a task which involves a lot of human labor.
In order to make the most out of the dormant potential in this huge collection of future-relevant knowledge, it is necessary to bring in software tools which can help the user to put in new insights, to link them together and to get out the desired results. The first tool in this direction is the extractor software which is now available in the Shaping Tomorrow website.
The extractor is a software service which works invisibly in the background and helps the user in two ways:
- It extracts metadata like publication date, author, source, country, region, keywords automatically from the input URL and adds it to the insight. By this it helps the user with the boring “house-keeping tasks” which either cost a lot of time or (understandably) are not done at all.
- It gives suggestions to the user about which parts of the texts are relevant for the future. This is the contents of the “Changes” and “Implications” fields.
It is very important to note that the output of the extractor is always meant as a first draft for the user who then can – and should – check and correct all fields before they are saved. This is especially important for the implications category:
- The software is in an early stage and still makes mistakes. But even in improved versions, the software cannot really “understand” a text on a deeper level. What it can do is to detect relevant text parts based on hints and clues in the text and suggest these to the user who has the final say.
- The software can only extract from the text what is in it. But some implications – and probably the more interesting ones – come from a combination of an insight and the user’s knowledge. For example a text may say that carbon emissions will rise in the next 20 years. But implications of that (temperature increase, sea level rise, etc.) might not be stated in this text, but come from the user’s knowledge about cause-effect relations in the physical word.
To summarize, the extractor is a tool which does not fully automate the entry of new insights into the Shaping Tomorrow knowledge base, but which should help the user to work more productively and is there to assist the user in thinking about changes and future implications. The extractor about halves manual capture time and increases the quality and consistency of human tagging.
In a follow-up blog entry I will describe the technology on which the extractor is built and after that I will write something about future goals for the extractor and other tools in order to bring out the full value of the knowledge stored inside Shaping Tomorrow’s collection of insights. I would also love to hear from users all kinds of feedback on the extractor; criticism, ideas, suggestions and whatever comes to mind.