Rosanna introduces the Adjacency Matrix to store the social network in order to fully record the connection between all the users in the corresponding social media. But, when it comes to dealing with the sparse network relation, a fully-filled adjacency matrix with 0 or 1 cannot save the storage space and then influence the time complexity when some needs about updating the whole matrix exist.
Maybe, a binary tree may help to save the statistics on social network. However, this problem is fairly different from that about seeking a shortest routine, when a distance network has been given.
Automatically, the original network database helps to solve this problem. To save the memory when to run a software, the relatively advisable method is that when needing statistics in the database, programmers just need to extract the corresponding data from the database.
I remember that Google have thought of a new data structure to store the net data and published a pdf to describe that. The pity is that I cannot find the material on it.
2014年10月17日 星期五
2014年10月2日 星期四
The link between data mining and graph memory
A large amount of information is created and stored by social medie, from which we can anglyse the preference of people towards something, such as, how they feel or evaluate. A method in data mining is called aspect-level analysis, which extracts the characaterised aspects or attributes of an entity. Here, I think, is fairly similar with the memorizing method of graph memory.
Take two-dimension code for example. Two-dimension code is invented before 1990 and afterwards in 2012 applied in large-scale among Chinese mobile market by Wechat. There are few people who can memorise the key information of two-dimension code so that they can distinguish one among a group of graphs containing two-dimension codes. The key that these people can realize this is that they extract the key aspects of this kind of code.
Now, let us think about,when we are in our childhood, how our parents teach us to recognize one person from so many people. Undoutedly, our parents teach us to focus on their color and length of hairs, whether wearing glasses, figure and weight, which all are the key-aspect of a person.
We memorize a graph or scene through catching the characters. Data mining is extremely like this.
Algorithms written as a few lines of codes or an application cannot behave well like humans upon extracting all the potential information of a few words. Therefore, what we can do is to analysing the key characters and digitalise these characters so that we can form a more direct image towards the information.
Take two-dimension code for example. Two-dimension code is invented before 1990 and afterwards in 2012 applied in large-scale among Chinese mobile market by Wechat. There are few people who can memorise the key information of two-dimension code so that they can distinguish one among a group of graphs containing two-dimension codes. The key that these people can realize this is that they extract the key aspects of this kind of code.
Now, let us think about,when we are in our childhood, how our parents teach us to recognize one person from so many people. Undoutedly, our parents teach us to focus on their color and length of hairs, whether wearing glasses, figure and weight, which all are the key-aspect of a person.
We memorize a graph or scene through catching the characters. Data mining is extremely like this.
Algorithms written as a few lines of codes or an application cannot behave well like humans upon extracting all the potential information of a few words. Therefore, what we can do is to analysing the key characters and digitalise these characters so that we can form a more direct image towards the information.
訂閱：
意見 (Atom)