Select a research journal that you believe is relevant for media technology research. The journal should be of high quality, with an “impact factor” of 1.0 or above. Write a short description of the journal and what kind of research it publishes.
Select a research paper that is of high quality and relevant for media technology research. The paper should have been published in a high quality journal, with an “impact factor” of 1.0 or above. Write a short summary of the paper and provide a critical examination of, for example, its aims, theoretical framing, research method, findings, analysis or implications.
From the Journal of Web Semantics, volume 9, issue 4 in December 2011, I have chosen the article "Semantically enhanced Information Retrieval: An ontology-based approach". The article tackles the issue with keyword-based search models, the way that we usually search for things using the web. One of the issues with keyword-based searching is that someone who wants to search for a specific topic might not know the terms needed to find what they need. A lot of research has been conducted regarding "conceptual search", understood as searching by meanings rather than literal strings which would make it easier to dive into an area you are unfamiliar with. The authors of the article takes a step further and proposed an ontology-based information retrieval model. Practically, the difference is that you could use a more natural way of creating your queries when retreiving information.
I find the content of the article very interesting, as using natural language when searching is getting more and more popular for each version of operating systems that get released. As a real world example, Apple's virtual helper Siri uses an ontology (among other methods as well) model to understand what it is you want to search for when you ask it "what is the weather going to be like tomorrow?". Search engines and virtual companions such as Siri, Google Now and Microsoft's Cortana are getting better and better each day at understanding natural language, using machine learning and ontology-based models.
The main focus of the article is to bridge the gap between the information retrieval and the semantic web communities in the understanding and realization of semantic search. This is great as it will help people interact with computers, to more easily find what they want. The computers have the information, we just need to have a natural way of retreiving it.
One problem with the evaluation is that when it comes to the semantic web, there are not any standardized evaluation techniques as compared to the information retreival community which means that there is not a defined way to judge the quality of semantic search methods. The authors have thus conducted their evaluations based on a couple of user-centred methods that are hard to recreate.
From the Journal of Web Semantics, volume 9, issue 4 in December 2011, I have chosen the article "Semantically enhanced Information Retrieval: An ontology-based approach". The article tackles the issue with keyword-based search models, the way that we usually search for things using the web. One of the issues with keyword-based searching is that someone who wants to search for a specific topic might not know the terms needed to find what they need. A lot of research has been conducted regarding "conceptual search", understood as searching by meanings rather than literal strings which would make it easier to dive into an area you are unfamiliar with. The authors of the article takes a step further and proposed an ontology-based information retrieval model. Practically, the difference is that you could use a more natural way of creating your queries when retreiving information.
I find the content of the article very interesting, as using natural language when searching is getting more and more popular for each version of operating systems that get released. As a real world example, Apple's virtual helper Siri uses an ontology (among other methods as well) model to understand what it is you want to search for when you ask it "what is the weather going to be like tomorrow?". Search engines and virtual companions such as Siri, Google Now and Microsoft's Cortana are getting better and better each day at understanding natural language, using machine learning and ontology-based models.
The main focus of the article is to bridge the gap between the information retrieval and the semantic web communities in the understanding and realization of semantic search. This is great as it will help people interact with computers, to more easily find what they want. The computers have the information, we just need to have a natural way of retreiving it.
One problem with the evaluation is that when it comes to the semantic web, there are not any standardized evaluation techniques as compared to the information retreival community which means that there is not a defined way to judge the quality of semantic search methods. The authors have thus conducted their evaluations based on a couple of user-centred methods that are hard to recreate.
Theory is a way to explain, describe or enhance the understanding of the world using an empirical model. This means that we can, by experimenting and trying out what works and what does not - explain and describe a phenomenon. We can also use these models, the theories, to try and look into the future, by looking at how things have previously worked. An example could be that by throwing a rock, we can see and understand that it will eventually hit the ground again - because of the way that gravity affects the rock. Now, we could also assume that an object of similar weight would also behave the same way! We would be able to foresee, and thus in a way look into the future, that an apple would also hit the ground if we threw it like we did the rock. By doing this, we have created a very basic theory about gravity, and how it affects everything we see around us by pulling it towards the ground.
To understand what a theory is not, we will take help from Sutton and this text What Theory is Not. Sutton brings up five points to describe things that are usually mistaken for theories:
- References - Could be used as background to explain something, but references by themselves are not theories.
- Data - We can do lots of tests, but only by explaining why the tests end like they do, we can create a theory.
- Lists of variables or conducts - Sort of like the "data" point. We need explanations!
- Diagrams - We can easily represent a theory with diagrams, but we need the whole story why a diagram looks like that to have a theory.
- Hypotheses/predictions - These are basically guesses, not theories. Hypotheses tell us what is expected to occur, not why it is expected to occur.
Describe the major theory or theories that are used in your selected paper. Which theory type (see Table 2 in Gregor) can the theory or theories be characterized as?
The theory in the paper is that an ontology-based model for information retrieval can be used for information retrieval. The theory could be characterized as an explanation and prediction, as the results from the evaluation shows that their method of a semantic algorithm is more efficient than the traditionally used ways they evaluated. I guess it could also be characterized as design and action, as the theory encompasses their semantic method, which they use to conduct queries.
Which are the benefits and limitations of using the selected theory or theories?
Is it easier to describe the phenomena in a paper, which makes it easier for further papers to be written about the same topic, and just makes the whole concept easier to understand by realizing the theory. By having an explanatory and predictive paper, it is very easy to see what the paper is about, and easy to understand what the outcome of the paper could be. It it quite binary - either their proposed method is better, or it is not. The limitations, I guess, would be that the testing may not always be the same. As the authors mentioned, the evaluation methods regarding this topic is not really standardized and it may be hard to recreate or compare this test with a new paper.
Inga kommentarer:
Skicka en kommentar