Imparting knowledge through storytelling, Part 2

That is a good and necessary trend, but the situation with metadata for stories is even less developed and requires some different thinking. One reason for that is the different structural and temporal elements of stories.

It is beyond the scope of this article to propose complete metadata schemata for stories, but I would like to look at three areas that could fruitfully be developed:

• metadata extensions--to capture the sort of narrative schemata that researchers have developed;
• XML within stories--particularly to capture the temporal relationships of stories; and
• RDF for representing the rich, multidimensional knowledge contained in stories.

There have been several proposed schemata that all stories follow. One describes stories as consisting of five stages revolving around equilibrium: equilibrium, disruption, recognition, effort to restore and results of effort to restore.

Another schema uses different labels and subdivides two of the stages and comes up with the following schema: introduction of setting and characters, explanation of state of affairs, initiating event, emotional response or statement of a goal by the protagonist, complicating actions, outcome and reaction to the outcome. If we look at those stages as metadata fields, we might fill them with the following types of values:

• introduction--description, names of characters, keywords or concepts, subject matter or category;
• explanation—description;
• initiating event--short description (between description and entity);
• emotional response—description;
• complicating actions--short description, keywords or concepts;
• outcome—description; and
• reaction--description, keywords as part of an ontology of lessons or morals of stories.

Having a set of metadata fields such as those could form the basis of an organizational scheme, but still would not capture the complex relationships found in stories.

XML

Another approach to the narrative schema above could be using XML to delineate the stages:
     <Introduction></Introduction>
     <Initital Explanation></Initial Explanation>
     <Initiating Event></Initialing Event>
and so on.

XML could also be used with similar story elements or relationships. For example, and tags could be used to relate the timing of two events. A similar relationship, though more dependent, might be the relationship. If one element exists in the story, then the preceding one must exist as well.

Other relationships are conflict<>resolution and <supports> and <opposes> .

RDF

Metadata is good for capturing certain aspects of the content of stories, and XML is particularly good at structuring the sequential nature of stories, but RDF might supply a missing element that is a way of capturing the rich conceptual relationships within stories. Those relationships include both subject matter hierarchies within which a concept or keyword is located and a typology of world relations.

A story that deals with lying, for example, with the right RDF description would also be related to broader concepts such as dishonesty and ethics. In other words, lying (is a member of> dishonesty and dishonesty <is a member of> ethics.

Adding a typology of world relations also helps provide a layer of structure that would greatly improve a user’s ability to find what they are looking for. In our lying example, if “lying” was a keyword that described one of the events in a story, the types of keywords that lying was related to might include things like the results of lying, the morality of lying, lying about subject matter X and so on.

Another important element of metadata for stories, whether expressed in RDF or some other format, is distinguishing between concepts or keywords that are merely mentioned and those that are elaborated on. This is an area that many of the automatic categorization companies are dealing with also. In that arena, it is referred to as the “aboutness” of a document, and there are different approaches to characterizing the aboutness of a document ranging from a weighted set of keywords to barcodes to a hierarchical representation of the concepts within a document.

There is no clear answer for how to create a metadata framework for stories, but one thing is clear and that is just the attempt to create one will have enormous benefits for knowledge architecture and knowledge retrieval. Stories provide a good test bed for dealing with the additional complexities of knowledge because they are relatively well structured and there is a rich body of existing and new material being generated all the time.

Future directions

It should be pretty clear from the above that we still have a long way to go to create a knowledge architecture that will support the full value of storytelling in business. I offer the following, not as a specific goal or approach, but as an image of one way it might work in the future.

The first point is that a knowledge architecture for stories must be much richer than traditional library or information architectures. That richness permeates the entire process, from the multiple contexts of stories themselves to the variety of categorization schemata that must be developed. In addition, both the representations of stories and the story categorization must be much richer than currently available. There are many ways to add that richness, and some current efforts point to particularly valuable approaches such as the use of collaboratively generated archetypes treated as