yes, it’s one of algorithm in my Thesis 
. Actually my topic is Knowledge Graph for Retrieval System, and first of all, I have to understand text summarization before starting this. Maybe Hilma could help me with some stuff that you have explored it before. I’d like to send you a message but your number is nowhere . It’s like a pain for people after change their cell phone, lose some phone numbers. If anyone of you would like to call Hilma, please tell her about it.
Well back to laptop :p, I think of how I could do semantic retrieval everyday, and make me headache, hikz. Finally I found it few days later, the method was using sample 
(inspired by Numb3rs, he3) I have not yet discuss it with my lecturer, so it was first solution for knowing the topic of the inserted document. Let me point of this, if I have a criminal document, the most subject you will be found are suspect, police, victim, murderer. (I do my algorithm for Indonesian document, because it has already done in English). Then the most verb you will be found are murder, rape, bust, arrest. After that, I make a program to link it in a graph, for example, police –> suspect is causal + (honestly, I’m not yet much understand about graph theory :”> since I didn’t join Graph Algorithm class )
Finally, after I make a link for some subjects, verbs, and objects, I’ll rate it. Maybe the database architect is like below.
Link1 Link2 Causality Rate
———————————–
Police Suspect Cau- 7.85
Police Murderer Cau- 8.12
Police Victim Cau+ 8.04
…………………………………………………..
…………………………………………………..
The rate was taken from
(Total relationship between link1 and link2 and Causality)/(Total Causalities)
It’s just a simple algorithm, and it looks trouble if the sample document is about 1000 and what I need is more than 1000 documents with different topics. It’s pain if I entry one by one relation in each document . (It looks gonna be failed)
The second alternative is rule based. I have to consult with the Indonesian lectures for it. It’s pain too since I’ve got C in my Indonesian class, oh noo . The latest alternative is using probability without using sample, and the question is, USING WHAT?
hahahaha,,, I was burst of laugh then. God, please help me, You are the smartest one in this world, and I know this problem is just piece of cake of you. There’s an answer but You’ll give me in the right times, I know that, so I keep pray to You… Ya Allah, please increase my think ability, please make my study easier, please make it smooth, Amin 
