Recommender systems

   This week we had learned a very interested notion recommendation.There is an extensive class of Web applications that involve predicting user responses to options. Such a facility is called a recommendation system.
   Firstly, I want to analysis why should we use such recommendation systems. On the one hand,there are values for the customer such as finding things that  are interesting,narrowing down the set of choices, helping us explore the space of options,discover new things,entertainment and so on . On the other hand,there are also advantages for providers such as Additional and probably unique personalized service for the customer, increasing trust and customer loyalty, Increasing sales, click trough rates, conversion etc,opportunities for promotion, persuasion. Obtaining more knowledge about customers and so on .

   Secondly,I notice  there are two basic architectures for a recommendation system:1.Content-Based systems focus on properties of items. Similarity of items is determined by measuring the similarity in properties2.CollaborativeFiltering systems focus on the relationship between users and items. Similarity of items is determined by the similarity of the ratings of those items by the users who have rated both items.In a contentbased system, we must construct for each item a profile, which is a record or collection of records representing important characteristics of that item. In simple cases, the profile consists of some characteristics of the item that are easily discovered.
   Lastly, I am going to introduce how to recommend items to users based on content.With profile vectors for both users and items, we can estimate the degree to which a user would prefer an item by computing the cosine distance between the user’s and item’s vectors.The random-hyperplane hashing techniques can be used to place (just) item profiles in buckets. In that way, given a user to whom wewant to recommend some items, we can apply the same two techniques – random hyperplanes and LSH – to determine in which buckets we must look for items that might have a small cosine distance from the user.

Graphs in social network analysis

  In the last week ,We learned a lot of new ideas in social media analysis.For me ,the use of graphs in SNA and Betweenness Centrality impressed me the most.Graphs are mathematical structures used to represent relationships between objects.In this way, we can view social network as graphs when doing social media analysis;Betweenness is a centrality measure based on shortest paths which is usually used in complex social network analysis.

  To begin with ,I would like to talk about graphs in social network.A social network is always represented as a sociogram,in which nodes represent users and edges represent the relationship between them. A sociogram is realizes by a graph as usual.Graphs offer us a way to describe and capture users’ behavior and interests in terms of their activities.In social network,there are different kinds of graphs,and a discussion graph is a hyper-graph representation of a set of relationships and their associated contexts, extracted from a social media corpus.A key advantage of using discussion graphs is that we can represent and analyse complex relationships.

   

  Then ,I would like to say something about betweennness centrality.When doing social media analysis, one of the fundamental thing is to determine importance(centrality) in a particular vertex(or an edge).Some of  the well-known methods are closeness,stress,and betweenness.Of these methods,betweenness is widely used in social analysis. Betweenness Centrality of a vertex is defined by the number of shortest paths in the network that pass through it.Consider a graph G(V,E),where V is the set of vertices representing actors or nodes in the complex network,and E ,the set of edges representing the relationships between the vertices.Teh number of vertices and edges are denoted by n and m respectively .The graphs can be directed or undirected.We assume that each edge has a positive weight W(e).For unweighted graphs,we use W(e)=1.A path from vertex s to t is defined as a sequence of edges.The length of a path is the sum of the weights of edges.Using Betweenness Centrality ,we can get to know the importance of each users in the network.

   

My thoughts on Opinion mining

  Microblogging nowadays has become a more and more popular way for people to communicate through the internet. Millions of messages are appealing daily in popular platforms such as sina weibo,twitter.People share their thoughts,feelings,opinions on various topic.
  Such data is not only useful for individuals but also for companies.On the one hand ,a lot of companies manufacturing such products have starter to poll these microblogs to get a sense of general sentiment for the product.Many time they analyse the data to study user reaction and reply to users.Here is an example for companies to do sentiment analysis:Twitter sentiment,a hedge fund named Derwent Capital Markets analyses the sentiment of uses of twitter to predict a stock.They focus on the feelings of a lot of users who are talking aboutstocks to help them.The picture blew illustrate the relationship between stock price and the emotion of twitter users:

  On the other hand,choices individuals make are often conditioned upon how others see and evaluate.Therefore,we often seek out the opinions of others when we need to make a decision.That is way we should do Sentiment Analysis and Opinion Mining.When mining data ,we usually make large quantities of feedback more manageable by splitting them into cluster based on keywords.Sentiment analysis,which is the second step,adds a dimension to the data.It makes it possible to focus the mining on areas in need of negative feedback or positive feedback.In this case ,we should use Sentiment Classification (polarity detection)to estimate the polarity of the opinions of the writer.The first approach should consist of paying special attention to words that tend to express a positive or negative attitude.However,this may not be straightforward as one may think,given that sentiment is often expressed in more indirect way.Therefore,we should study more deeper in order to deal with this case.
  As studying deeper and deeper ,I find what we learned in the class is really practical and useful.I will try to get a good command of it.