Actual Paper: http://www.islab.ntua.gr/attachments/article/63/06033365.pdf Link predictions is done as a classification problem for a competition. 94 features were calculated and they are given as input for Random forests for classification. ROC is used for evaluation. Study of network dynamics has numerous applications like marketers use it to recommend products to a person based on his contacts. […]
Monthly Archive for April, 2013
Graph-based Features for Supervised Link Prediction
Posted in Uncategorized on Apr 24th, 2013
Scalable Proximity Estimation and Link Prediction in Online Social Networks
Posted in Link Prediction on Apr 24th, 2013
Actual Paper: http://www.cs.utexas.edu/~yzhang/papers/osn-imc09.pdf Estimating proximity between nodes in social networks is challenging due to size and dynamic nature of networks. Proximity measure is the central measure in analysis of social networks. Proximity sketching and Proximity embedding are proposed in this paper. They are demonstrated using Link Prediction problem on multiple networks. Simplest proximity estimations are […]
SNAP – Stanford Network Analysis Project
Posted in Tools on Apr 17th, 2013
SNAP is a tool for network analysis and graph mining. It is implemented in C++ and can scale up to millions of nodes, with efficient data structures to optimize memory. The source code has lot of examples that will explain how to implement or traverse over trees / graphs using the tool. Many commonly used […]
Image Labeling on a Network: Using Social-Network Metadata for Image Classification
Posted in Link Prediction on Apr 17th, 2013
Original Paper: http://i.stanford.edu/~julian/pdfs/eccv2012.pdf This paper tries to predict tags, labels or recommendations for images based on social network meta data of the images, i.e., the connection between images, like images uploaded to same groups. The task is a binary labeling problem using structure learning techniques Challenge is to find how to effectively use/model social network […]
Supervised Random Walks: Predicting and Recommending Links in Social Networks
Posted in Link Prediction on Apr 17th, 2013
Actual Paper: http://arxiv.org/pdf/1011.4071.pdf Citiation: Backstrom, Lars and Leskovec, Jure Supervised random walks: predicting and recommending links in social networks WSDM ’11 The link prediction problem is to identify any links that may be formed in future, or missing links in a network. The challenge is to combine network structure and node/edge data. This problem is […]
Project problem statement
Posted in Project Specific on Apr 17th, 2013
Project Problem statement: We have a dataset of photos with tags, userids, location and label as features/metadata/dimensions. We will be creating a graph network of the photos based on common attributes, for example similar location and user id would link (create an edge) between two photos. We have the dataset with metadata as features and […]
Link Prediction using Supervised Learning
Posted in Link Prediction on Apr 16th, 2013
Actual Paper : http://www.siam.org/meetings/sdm06/workproceed/Link%20Analysis/12.pdf Below is a summary, understanding and takeaway of the paper. Link Prediction problem is to predict the likelihood of two nodes sharing an edge in future, given the current state of the graph. Co-authorship data set was used in this paper. This dataset also obey power-law distribution. The paper explains the […]