4 ms·

Sure you can compare it to a Wiki/CMS in some way. Importantly Substance is a Semantic Editor, that means you're editing Content, not the visual representation.

by _mql 16y ago

Sure you can compare it to a Wiki/CMS in some way. Importantly Substance is a Semantic Editor, that means you're editing Content, not the visual representation. Documents are expressed as data (as far as possible).

limmeau 16y ago
What is the document model behind it?
anigbrowl 16y ago
I think I get what you're working towards and will keep an eye on it. It will definitely attract more users and developers if you can highlight more use cases, and show where it differs from other platforms. When I see 'semantic' in relation to the web, right now I am often disappointed because it often involves an incomplete vision of 'the web would be more useful if everything was tagged properly.' What I want from a 'semantic' tool - and this is asking a great deal, almost science fiction - is a distributed ontology matcher with a basic ability to answer simple questions. This is a general challenge for the 'semantic web,' not just for you. So I hope you will keep exploring the possibilities. Let me give you an extreme example. I navigate to a news page and see 'Breaking: important event takes place!' Apparently there are 3000 news stories about this important event. Obviously, I do not want to read all 3000, and that number is only a reflection of how many people consider the story interesting. 2900 of those stories will probably be identical, give or take a few words. On a true 'semantic web' I would get an abstraction of the information common to all the stories, and tools to help me find the very small number of stories with unusual additional information - while filtering out websites run by fanatics and crazies. So in a semantic editor, I would like to write in natural language, and have the computer extract simple ontological fragments for use as complex search input. As I write about a particular subject, the computer looks at what I am writing about, and a constantly evolving selection of relevant reference material is available in a side panel. So if I begin writing about revolution in Egypt, I should begin to see more and more reference material about Egypt, history, political theory, and related subjects. I don't want to search for data; I want the computer to search out data continually as a background task, based on what I am writing about. You know those dense academic papers, where almost every sentence has a footnote even if it stating something very obvious and well-known (mainly in order for other readers to assess the quality of research)? A true semantic editor will narrow the gap between collection and composition. I am not expecting to just have this delivered, of course :) Instead, I hope it provides some stimulating ideas.
- _mql 16y ago
  Great pointers! What I really want to include in the nearer future is entity extraction. That means that we could send the text-contents of a document to an entity-extraction service (like OpenCalais) in order to get back a list of entities that are mentioned in the text (that's what documentcloud.org is doing). By doing this, not only our documents could be tagged automatically, but also scenarios like the ones you described are made possible (given that entity-extraction can be performed fast enough). What do you think? I may get back to you when I start working on it. It's on the queue for sure. :)

by _mql 16y ago

limmeau 16y ago

What is the document model behind it?

anigbrowl 16y ago

I think I get what you're working towards and will keep an eye on it. It will definitely attract more users and developers if you can highlight more use cases, and show where it differs from other platforms. When I see 'semantic' in relation to the web, right now I am often disappointed because it often involves an incomplete vision of 'the web would be more useful if everything was tagged properly.' What I want from a 'semantic' tool - and this is asking a great deal, almost science fiction - is a distributed ontology matcher with a basic ability to answer simple questions. This is a general challenge for the 'semantic web,' not just for you. So I hope you will keep exploring the possibilities. Let me give you an extreme example. I navigate to a news page and see 'Breaking: important event takes place!' Apparently there are 3000 news stories about this important event. Obviously, I do not want to read all 3000, and that number is only a reflection of how many people consider the story interesting. 2900 of those stories will probably be identical, give or take a few words. On a true 'semantic web' I would get an abstraction of the information common to all the stories, and tools to help me find the very small number of stories with unusual additional information - while filtering out websites run by fanatics and crazies. So in a semantic editor, I would like to write in natural language, and have the computer extract simple ontological fragments for use as complex search input. As I write about a particular subject, the computer looks at what I am writing about, and a constantly evolving selection of relevant reference material is available in a side panel. So if I begin writing about revolution in Egypt, I should begin to see more and more reference material about Egypt, history, political theory, and related subjects. I don't want to search for data; I want the computer to search out data continually as a background task, based on what I am writing about. You know those dense academic papers, where almost every sentence has a footnote even if it stating something very obvious and well-known (mainly in order for other readers to assess the quality of research)? A true semantic editor will narrow the gap between collection and composition. I am not expecting to just have this delivered, of course :) Instead, I hope it provides some stimulating ideas.

_mql 16y ago
Great pointers! What I really want to include in the nearer future is entity extraction. That means that we could send the text-contents of a document to an entity-extraction service (like OpenCalais) in order to get back a list of entities that are mentioned in the text (that's what documentcloud.org is doing). By doing this, not only our documents could be tagged automatically, but also scenarios like the ones you described are made possible (given that entity-extraction can be performed fast enough). What do you think? I may get back to you when I start working on it. It's on the queue for sure. :)