Currently, employed to detect the posts containing terrorist propaganda. SNP filter, reads the social media posts and filters the results based on three specifications:

• Terrorism Tags
• Motivation Tags
• Out-of-context (OOC) Tags

Sample content of each Tag set is given in the table below:

Input: twitter
Output: preprocessedTwitter
1) Split the post into independent sentences 
2) Remove the stop words from text
3) Decompose compound data, such as #supportISIS =>support ISIS
3) Stem the words to root words, such as support, supported, supporting => support

Input : a new twitter
Out put : Accept or not
1)if the twitter contains words in filter set 1 then 
2)   if the twitter contains words in filter set 2 then
3)        if the twitter contains words in filter set 3 then
4)               return not accepted
5)        else
6)               return accepted 

The code in Mainapp.java is doing basic filtering. Run this file, in the result, should see a set of filtered twitter set in the standard output. This set should contain the twitters that have motivational terrorism information in them.

Example of multiple nouns : Suiside Belt, cut their heads off, call upon

Since our matching algorithm workds via key word matching in a 1-gram model, such that it can't work well on Multiple Nouns. For example, it will be hard to filter out a twitter that contains the object Suicide Belt. The problem happens because when we break down the sencenten, the smallest unit after dividing is a token made up of one word, such as embasy, isis, division etc. In the meantime, the tags sets are also made up of single word, such as terrorism, isis, threaten.