Loading...
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 45233 (19)
- University: Sharif University of Technology
- Department: Computer Engineering
- Advisor(s): Ghodsi, Mohammad
- Abstract:
- In many network which relationships between nodes are defined based on the similarity of attributes (Such as the World Wide Web and social networks), extracting information about networks object’s attributes may be difficult or even in many cases impossible.In these cases, predicting unknown attributes based on other objects attributes according to network structure can be extremely useful.Even more, finding a representative sample set of objects and trying to obtain their attributes in order to predict other object’s attributes with this obtained data can be an interesting problem. Finding such a set of objects with minimum size while giving maximizing accuracy in predicting other object’s attributes, can decrease data collection cost dramatically.
In this thesis, a global optimization algorithm is proposed based on minimizing neighbour’s attribute value difference to predict unknown attributes. Then the concept of similarly tendency is defined as the tendency of an individual object in the network to make relations with similar objects and used as an inherent fact in many networks to futrher improve the prediction algorithm.Then an iterative method is proposed for finding the representative sample set. In this method, using similatly tendency concept, in an iterative manner,objects which maximize out knowledge about network are selected and their information are used for predicting other objects attribute values and selecting other objects - Keywords:
- Network ; Iteration Method ; Representative Sample Set ; Similarity Compulation Model
Digital Object List-
محتواي کتاب - view
Bookmark