The limits of Web metadata, and beyond
read more
Citations
Trawling the Web for emerging cyber-communities
Shaping the Web: Why the Politics of Search Engines Matters
Toward a basic framework for webometrics
Influence and passivity in social media
ScentTrails: Integrating browsing and searching on the Web
References
Fuzzy models—What are they, and why? [Editorial]
The quest for correct information on the Web: hyper search engines
Metadata for the Masses
Rating the Net
Related Papers (5)
Frequently Asked Questions (9)
Q2. What is the method for enhancing a region of the Web?
For their general experimentation the authors have employed as metadata classification the well known Excite Ontology, also known as the "Channels Classification": it is a tree-like set of attributes (also known as categories), and is one of the best existing metadata sets for the general classification of World Wide Web objects.
Q3. What is the meaning of a certain set of metadata?
The "judgment" of a certain set of metadata with respect to a certain Web object consists in a number ranging from 0 to 100, with the intended meaning that 0 stands for a completely wrong/useless metadata, while 100 for a careful and precise description of the considered Web object.
Q4. How many Web objects were used in the first round of tests?
The first round of tests was done setting the size n of the Web region to 200, and the number m of Web objects that the users had to classify to 20 (which corresponds to the 10% of the total considered Web objects).
Q5. What is the way to improve the situation of the World Wide Web?
It has been realized recently that the only feasible way to radically improve the situation is to add to Web objects a metadata classification, that is to say partially passing the task of classifying the content of Web objects from search engines and repositories to the users who are building and maintaining such objects.
Q6. how many objects are classified in the web?
When employing a particular metadata classification for the Web, even when the percentage of classified Web objects is relatively small, usage of the back-propagation method can significantly help the effectiveness of the classification, thus helping the metadata to get more and more widespread, especially in the initial crucial phases when the number of classified objects will be extremely limited.
Q7. What is the way to measure the effectiveness of the approach?
In order to measure the effectiveness of the approach, the authors first need to "restrict" in some sense the Web to a more manageable size, that is to say to perform their studies in a reasonably sized region of it.
Q8. How do the authors add a region of the Web to the set of objects?
2.Once a region S of the Web has been selected this way, the authors have to randomly select a certain percentage p, and manually classify them with metadata; then, the authors propagate the metadata using the back-propagation method.
Q9. How many basic concepts can be kept reasonably small?
Indeed the number of basic concepts can be kept reasonably small, and then fuzzification can be employed to obtain a more complete and detailed variety of descriptions.