NxFilter Tutorial
Tutorial Index

GUI - Classifier
This is for NxClassifier that is the auto-classification engine for Jahaslist. It does dynamic classification against the websites visited by your users based on a keyword matching and scoring system. You can define or modify its classification ruleset as you like.


Classifier > Setup
- DNS Test Timeout : NxClassifier only classifies the existing domains. So it does DNS testing first.

- HTTP Connection Timeout : After DNS testing, now it needs to download a webpage to analyze. This is the connection timeout value for HTTP connection.

- HTTP Read Timeout : This is the data read timeout value after you have an HTTP connection.

- Classified Data Retention Days : NxClassifier keeps the classification result log for the recently classified domains. NxClassifier doesn't do the classification against these already classified domains.

- Keep HTML Text : NxClassifier extracts text from the first page of a website and keep it for reclassification. But this requires more disk space so you can decide to keep the text or not.

- Disable Domain Pattern Analyzer : NxFilter has a domain calssification process based on domain patterns. If a domain can be classified by this domain pattern analyzer NxClassifier doesn't try to classify the domain by other methods.

- Disable Cloud Classification : When NxClassifier fails to classify a domain, Jahastech's cloud based classifier will try to classify it.

- Disable Classification : You can disable the classification if you want.


Classifier > Ruleset

You need to understand how to make a classification rule. A classification rule consists of the following parts.

- Keyword : Matching keyword. In reality, it is a regular expression.

- Target : You can apply your keyword against the domain, title, description and text of a website.

We get the title, description and text of a website's first page.

- Points : You can set a different points to a rule by its importance. The minimum points to be classified is 100 and the maximum points is 1,000.

- Category : Associated category to a classification rule.

When you want to exclude a keyword for a category, set a negative number as the points for the keyword.


Classifier > Classified
This is the classification result log by NxClassifier. It will show you the recently classified domains and how they are classified or unclassified. Based on this classification result, you can improve your classification ruleset and you also can reclassify the already classfied domains here. To reclassify the domains already classified, use 'RECLASSIFY ALL' button.

With 'TEST' button, you do a test run for the current ruleset against a domain.


Classifier > Excluded
We exclude the domains making certain errors during the classification process. For example, if we have 403 response from a website we don't need to try to classify it as we can't access the website. Or if we get an image file or some other type of file instead of a text or HTML file we will exclude it.

We don't delete these excluded domains. If you want to let NxClassifier try to classify an excluded domain, you need to delete it from the list first.


Classifier > Blocklist
You can download and merge the public blocklists from the Internet into Jahaslist and Globlist overnight automatically on 'Classifier > Blocklist'.

Globlist is a part of Jahaslist.

- Format of the blocklist : Host file format or domains separated by new lines. Basically, all the blocklists from https://firebog.net will be working.

- Priority Points : There may be duplicated domains in your blocklists associated to different categories. You can make a specific blocklist to be downloaded and merged before others by setting higher priority points.

When you delete a blocklist URL, the domains merged from the blocklist will be lost. If you just want to exclude a blocklist from the overnight merging process, set its priority points to -1.

However, we don't merge every domain from your blocklists into Jahaslist. There are many false positives and non-existent domains in these public blocklists so we exclude some domains by the following rules.

1. If it's a non-existent domain.

2. If it's already in Jahaslist.

3. If it's in 100,000 well known domain list.

We may add domains already classified by Jahaslist as we only do 'Exact Matching' rather than 'Parent Domain Matching' in the merging process. This is for faster processing and not having performance impact on your system.


Classifier > Jahaslist
You can view the contents of Jahaslist and modify it directly here. But we don't recommend you to do the reclassification here unless it is a mass importation of domains. We keep Jahaslist in a separated DB file and NxFilter doesn't do auto-backup for it. So it is better to do it on 'Category > System' as the reclassified domains will be stored into the main config DB.


Classifier > Test Run
After you add your own classification rules, you want to see how effective they are. You can do a test run for your classification ruleset against a website here.

'Test Run' doesn't do actual classification.