1. How can we make better use of tags?
With eTag vision pretty defined, another question comes up, how can we make better use of these tags consumers collaborate.
Tagging systems use descriptive keywords to mark up webpages for future navigation, filter and search. eTag system allows user to collaboratively contribute to mark up images. Collaboratively tagging offers an interesting approach for classifying resources on the web comparing with conventional web ontology. It allows almost everyone – especially consumers – to freely attach keywords or tags on resource. This is most useful in the scenario that there’re too much information to classify. In eTag context, we maintain the tags of images which results in ranking system and recommendation system.
Considering a typical user of eTag system, as the number of tags he or she uses increase, we might find pattern regarding to how he marks up image by calculating the intersection of his tags with the rest of users’ tags (for filtering) and union (for recommendation). What’s more,as tagging is neither exclusive nor hierarchical and therefore can be used combined with conventional ontology to work out a very interesting and accurate image recommendation system.
People may feel interesting of some of the favourite tags. As we can easily find out which tags are popular and latest. Then we can recommend users to have a look.
2. Difficulty for eTag
There’re three problems in making our tags useful: polysemy, synonymy, and basic level variations.
A polysemous word is one that has many (“poly”) related senses(“semy”). For example, a “window” may refer to a hole in thewall, or to the pane of glass that resides within it . In practice, polysemy dilutes query results by returning related but potentially inapplicable items. As users expects accurate results, this may result in unwanted filter or search results.
Synonymy, or multiple words having the same or closely related meanings, presents a greater problem for tagging systems because inconsistency among the terms used in tagging can make it very difficult for one to be sure that all the relevant items have been found. For example, if television is searched, TV could be neglected. Another case is the plural form of words and spelling differences. For example, if dog is searched, dogs could be neglected or if colour is searched, color could be neglected.
The basic level variation closely relates to user’s expertise. For example, for a C++ programmer, when he searches C++, he might expect some very specialised C++ articles returned. But for ones don’t know about programming, he might want to just read some basic background knowledge of C++. This causes inflict.
3. What to do next?
I’ll dig more into currently working tagging system such as delicious and some app in popular social networking sites. This helps us to justify how mature tagging systems perform in reality.
I’m sorry I am in hospital all the time this week, I’ll help Jonny with the database design diagram and wireframe.
4. An interesting idea
Well, during literature search, I came across a paper, in this paper the author mentions a cheeky idea, but I think it’s very interesting. In a website, you can describe what kind of girl you like and the system will search popular social networking sites for you and find your dream girl, you can register and talk to her then. Good luck:). Lol.
5. Question:
Decentralization: The Future of Online Social Networking – I dont quite get the idea. What’s the difference between facebook and decentralization of the social networking sites.
Answer by Jonny: Current social networks store all of your data in closed off worlds, separate from the rest of the internet. You need to be a member of those social networks to see this data, and this means innovation is stifled because the only people really able to innovate are Facebook/Twitter/LinkedIn/etc.
An early (failed) example of this strategy was AOL. At the time, for many users the internet was AOL and they didn’t know about the many interesting websites outside of those walls. A lot of non-technical users now rely on facebook for everything – subscribing to news feeds through facebook pages instead of using RSS, etc.
AOL failed because the walled garden strategy can’t really innovate as quickly as an open system. Eventually, Something will come along to replace Facebook (as the web replaced AOL), and a distributed image tagging service is going to be part of that.
Decentralisation means that no single body controls everything – in my proposed model of etags it would mean anyone could start up a competing image service without effecting anyone else and people could move to that one, still get updates from their friends, etc. without much hassle.
I thought the more users tag one content, the more helpful for the accuracy of the recommendation system, but research done on delicious, the more tags, the worse accurate. It’s strange.
6. articles read this week.
The Structure of Collaborative Tagging Systems
Long Live the Web.
Understanding navigability of social tagging systems
Linked Data – the story so far