responsibility-commit Mailing List for Net Responsibility (Page 4)

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Revision: 320
          http://responsibility.svn.sourceforge.net/responsibility/?rev=320&view=rev
Author:   dustsource
Date:     2012-06-03 01:35:39 +0000 (Sun, 03 Jun 2012)
Log Message:
-----------
Made the following edits to the documentation related to rating improvement data:
--Added the "Rating Improvement Data" guide.
--In the "Configuration Explained" guide:
	--'Configuration Parts' section, 'Tweaks', 'Provide improvement data': Clarified relationship between this setting and rating warnings in reports.
--In the "Understanding the Reports" guide:
	--'Improving the reports' section:
		--In the last paragraph (item), clarified relationship between the 'Provide improvement data' setting and rating warnings in reports.
		--Added a final item linking to new "Rating Improvement Data" guide.
--In the "Troubleshooting the Reports" guide:
	--'Why are there so many false positives?' section:
		--Adjusted the text to help clarify relationship between the 'Provide improvement data' setting and rating warnings in reports.
		--Added brief comment about and link to the new "Rating Improvement Data" guide.
--In the "Privacy Policy" document:
	--'What about the improvement data?' section: Clarified relationship between the 'Provide improvement data' setting and rating warnings in reports.

Modified Paths:
--------------
    documentation/pages/0301-ConfigurationExplained.page
    documentation/pages/0705-UnderstandingReports.page
    documentation/pages/0709-TroubleshootingReports.page
    documentation/pages/0910-PrivacyPolicy.page

Added Paths:
-----------
    documentation/pages/5050-RateImprovData.page

Modified: documentation/pages/0301-ConfigurationExplained.page
===================================================================

--- documentation/pages/0301-ConfigurationExplained.page	2012-05-27 00:52:27 UTC (rev 319)
+++ documentation/pages/0301-ConfigurationExplained.page	2012-06-03 01:35:39 UTC (rev 320)
@@ -27,7 +27,7 @@
   - <*Instant threshold*>: Here you can enable instant reports, which means your accountability partner will get a report instantly when NR thinks you're looking at inappropriate content. You can also set the sensitivity. Entering zero means this feature is disabled. Otherwise, lower numbers mean higher sensitivity. (0 = Disabled; 1-5 = Very sensitive; 6-20 = Medium; 21+ = Low sensitivity)
   - <*Report strength threshold*>: Only matches with a strength higher than this number will be included in the <*Warnings*> section of the reports. That way you're able to set some sort of sensitivity for what to display. This number ranking system for these items is a fairly new feature that still needs to be tested and tweaked in order to make it accurate and useful. To find out more about this ranking system, including how you can help in making it more accurate, see [[0705-UnderstandingReports.html#4 this section]] of the "Understanding the Reports" guide.
   - <*Always perform tokenmatch*>: Using tokenmatch increases the report creation time because it's more complicated, but it results in much more accurate reports. If you're having problems with reports taking too long to create you might want to exclude it.
-  - <*Provide improvement data*>: Use this option if you'd like to help us with the development of NR. The program will <*anonymously*> send your warnings to our server and store them there. This will help us see how to improve the blacklists.
+  - <*Provide improvement data*>: Use this option if you'd like to help with the development of NR. The program will <*anonymously*> send your warnings to our server and store them there. This will help the developers see how to improve the blacklists. (Note that if you or your accountability partners choose to [[0705-UnderstandingReports.html#9 help improve the accuracy of reports]], the improvement data related to any rated warnings will be sent to and stored on the NR server, even if this option is not selected in your settings.)
   - <*Send me newsletters*>: Choose this if you want the latest information about new versions, how to help, and other things that might be of interest. These emails are not sent on a regular basis, but rather whenever there's a need to update users with fresh information.
 
 !!Blacklist categories

Modified: documentation/pages/0705-UnderstandingReports.page
===================================================================
--- documentation/pages/0705-UnderstandingReports.page	2012-05-27 00:52:27 UTC (rev 319)
+++ documentation/pages/0705-UnderstandingReports.page	2012-06-03 01:35:39 UTC (rev 320)
@@ -80,4 +80,5 @@
 %>
 
 In the drop-down menu, the first number is simply the current ranking that NR has given this item. The numbers after that (0, 25, 50, 75, 100) represent various points or markers along the 0-100 continuum that represent levels of the likelihood of an URL being a cause for concern or not. See the [[#4 Warnings]] section on this page to see a description of what each of these levels is meant to represent.
-  - So the way to help out is to look at each item, evaluate which of the levels it best fits into (according to the above-mentioned description), and then choose the corresponding number from the dropdown menu. When you do this, your rating of the URL will be sent to the NR server, which will then help the developers know how to make any necessary adjustments to NR's filtering methods. (Once you have rated an URL, the number you chose will show up on a white background and the drop-down menu will no longer be available for that item. However, if you close and re-open the report, it will appear as it did at first.) Note that it is helpful to assign a rating to an URL regardless of whether it's current ranking is accurate or whether it is way off. Also, the more people that are regularly reporting ratings of URLs in this way, the more accurate the reports will become. So your help would definitely be appreciated!
+  - So the way to help out is to look at each item, evaluate which of the levels it best fits into (according to the above-mentioned description), and then choose the corresponding number from the dropdown menu. When you do this, the URL and its related information, along with your rating of it, will be sent to the NR server. (Note that this information is then stored on the NR server even if the user has not selected 'Provide improvement data' in their online settings.) This will then help the developers know how to make any necessary adjustments to NR's filtering methods. (Once you have rated an URL, the number you chose will show up on a white background and the drop-down menu will no longer be available for that item. However, if you close and re-open the report, it will appear as it did at first.) Note that it is helpful to assign a rating to an URL regardless of whether it's current ranking is accurate or whether it is way off. Also, the more people that are regularly reporting ratings of URLs in this way, the more accurate the reports will become. So your help would definitely be appreciated!
+  - If you would like to help improve the accuracy of reports by rating improvement data provided by other users as well, see the [[5050-RateImprovData.html "Rating Improvement Data" guide]] for more information.

Modified: documentation/pages/0709-TroubleshootingReports.page
===================================================================
--- documentation/pages/0709-TroubleshootingReports.page	2012-05-27 00:52:27 UTC (rev 319)
+++ documentation/pages/0709-TroubleshootingReports.page	2012-06-03 01:35:39 UTC (rev 320)
@@ -18,6 +18,7 @@
   - The attached report may be 'broken' because of some problems in its code (such as the appearance of unwanted newlines and single quotes). If this is the case, the usual symptom is that JavaScript will not work at all, meaning that neither the JavaScript tree nor the text message about JavaScript being disabled will be displayed. Improvements to the software are the best long-term solution, but until that happens a temporary workaround is to try to 'repair' the broken report by clicking on [[http://netresponsibility.com/repair.php this link]] and following the instructions there. This is not guaranteed to work in every case, but it is worth trying. Also, note that the report you are trying to repair needs to be smaller than 12 MB.
 
 !!!Why are there so many false positives?
-There may be a lot of false positives (i.e. a warning that is not really a problem) in the <*Warnings*> section of the reports you receive, causing you to wonder why there are so many, and if there isn't anything we can do to lessen or completely avoid them. The reason there are so many is because we are still working at testing and tweaking our filtering method in order to catch what needs to be caught, and ignore the rest. The developers of NR can do a certain amount of improvement on their own, but it would be tremendously helpful to us if NR users and accountability partners also helped out with this process. Here are the two main ways that help can be given:
-  - Make sure the NR user has selected the option 'Provide improvement data' in the 'Tweaks' section of their online settings. When this is selected, the program will <*anonymously*> send the warnings from each report to our server and store them there. This will help us see how to improve the blacklists.
-  - A relatively new feature we are using in the <*Warnings*> section of reports is a color-coded number ranking system for each item listed. Because it has not been extensively tested, you should not completely rely on it at the current time. However, we are working on improving its accuracy and the good news is, since we're including it in the reports that are sent out, you are also able to help out with this. For more information about how you can give valuable help by giving a bit of your time whenever you receive a report, see the [[0705-UnderstandingReports.html#9 Improving the Reports]] section of the "Understanding the Reports" guide. Thanks in advance for whatever help you can give in making the reports more accurate and useful!
+There may be a lot of false positives (i.e. a warning that is not really a problem) in the <*Warnings*> section of the reports you receive, causing you to wonder why there are so many, and if there isn't anything that can be done to lessen or completely avoid them. The reason there are so many is because efforts are still being made to test and tweak the filtering method in order to catch what needs to be caught, and ignore the rest. The developers of NR can do a certain amount of improvement on their own, but it would be tremendously helpful if NR users and accountability partners also helped out with this process. Here are three different ways that help can be given:
+  - NR users can make sure the option 'Provide improvement data' is selected in the 'Tweaks' section of their online settings. When this is selected, the program will <*anonymously*> send the warnings from each report to our server and store them there. This will help developers understand how to improve the blacklists and filtering methods.
+  - A relatively new feature we are using in the <*Warnings*> section of reports is a color-coded number ranking system for each item listed. Because it has not been extensively tested, you should not completely rely on it at the current time. However, efforts are being made to improve its accuracy and the good news is, since it is included in the reports that are sent out, you are also able to help out with this. For more information about how you can give valuable help by giving a bit of your time whenever you receive a report, see the [[0705-UnderstandingReports.html#9 'Improving the Reports' section of the "Understanding the Reports" guide]]. Thanks in advance for whatever help you can give in making the reports more accurate and useful!
+  - You may also help improve the accuracy of reports by rating improvement data provided by other users as well. See the [[5050-RateImprovData.html "Rating Improvement Data" guide]] for more information about how you can do this.

Modified: documentation/pages/0910-PrivacyPolicy.page
===================================================================
--- documentation/pages/0910-PrivacyPolicy.page	2012-05-27 00:52:27 UTC (rev 319)
+++ documentation/pages/0910-PrivacyPolicy.page	2012-06-03 01:35:39 UTC (rev 320)
@@ -13,7 +13,7 @@
 
 !!!What about the improvement data?
 
-If you choose to send improvement data, it means all visited sites that are flagged as inappropriate are sent to our server. We do not store any information about who visited these sites, such as username, IP number, MAC address etc. This anonymous data helps us develop better filtering algorithms that in return will produce better reports.
+If you choose to send improvement data, it means all visited sites that are flagged as inappropriate are sent to our server. If users or their accountability partners choose to [[0705-UnderstandingReports.html#9 help improve the accuracy of reports]] by rating URLs NR has flagged, the data related to these URLs will be sent to the NR server, even if the user did not select the 'Provide improvement data' option in their settings. We do not store any information about who visited these sites, such as username, IP number, MAC address etc. This anonymous data helps us develop better filtering algorithms that in return will produce better reports.
 
 
 !!!What information do we collect?

Added: documentation/pages/5050-RateImprovData.page
===================================================================
--- documentation/pages/5050-RateImprovData.page	                        (rev 0)
+++ documentation/pages/5050-RateImprovData.page	2012-06-03 01:35:39 UTC (rev 320)
@@ -0,0 +1,110 @@
+Rating improvement data
+A50NR_DevIntro
+
+!!!Introduction
+There are currently three ways that Net Responsibility (NR) users and their accountability partners (APs) can help improve the accuracy of the <*Warnings*> section of the reports. One way is for the user to make sure they have selected the option 'Provide improvement data' in the <*Tweaks*> section of their [[0301-ConfigurationExplained.html online configuration settings]], which automatically and <*anonymously*> sends all <*Warnings*> items to the NR server where they can then be rated, analyzed, etc. in order to improve NR's filtering methods. A second way is for users and APs to [[0705-UnderstandingReports.html#9 rate individual items]] from the <*Warnings*> section of the reports they receive. This guide describes the third way you can help out, which is by rating the improvement data that has been received from multiple NR users. Please note that, while this is a great way to help out, you also need to be careful in how you go about it, since it is possible that while you are doing this you will see keywords and URLs that will tempt your mind to go in directions you don't want it to. In a later section of this guide, you will find some ideas and tips for protecting yourself as needed.
+
+All of the improvement data can be viewed and rated by using an online interface hosted on the NR server. [[http://www.netresponsibility.com/dev/improvement_data.php Click here to see what it looks like or begin using it.]] You may find it helpful to have it open in your browser as you read through the rest of this guide, which explains how it works, along with the various options that can be used with it.
+
+
+!!!Online interface options
+This section explains the various options available for you to customize your approach to rating the improvement data.
+
+!!The 'Where' field
+The warnings are loaded in a MySQL string, and if you are familiar with that syntax it will be easy to load specific types of warnings you're looking for. A percentage sign (%) will match anything or nothing. You may leave this field empty to load all warnings. Here are some examples of how to use SQL syntax in the 'Where' field to load warnings selectively:
+  - category LIKE 'porn'
+  - category LIKE 'personal blacklist'
+  - keyword LIKE 'sex%'
+  - strength < 50
+  - rating = 0
+  - category LIKE 'porn' AND strength < 50
+
+!!'Order by'
+Here you can choose the order in which you want the warnings to be shown. The options may be self-explanatory, especially if you've used a similar sort of sorting or ordering function before. This option can be combined with the options 'Ascending' or 'Descending' in order to reverse the order. Here is a list (same as in the drop-down menu) of the ways that the warnings can be sorted/ordered:
+  - Date: This is the default option, and refers to the date the warning was flagged. Combined with 'Ascending' it will show the earliest warnings first. This is helpful if you want to focus on the newest or oldest warnings.
+  - Amount: This refers to the number of times that this identical URL and a particular keyword match was sent to the NR server. Combined with 'Ascending' it will show URLs with the least amount first. This is helpful if you want to focus on rating URLs that are more frequent or less frequent.
+  - Rating: This refers to a number from 0 to 5, which represents the average of how a warning has been rated by users and/or accountability partners. If the number is 0, it means the warning has not yet been rated. The numbers 1, 2, 3, 4, and 5 represent points along a continuum from something that was definitely a false positive (lower number) to something that was definitely an inappropriate site (higher number). Because these are based on an average, you may see numbers with decimals like 1.23 or 3.45, etc., but they will all be within the 1-5 range if they have been rated before. This is helpful if you want to focus on warnings that are currently unrated or warnings that other users have considered to be more or less likely to be inappropriate sites.
+  - Votes: This refers to the number of times that this warning has been rated. This is helpful if you want to focus on warnings that have been rated less or more times by other people.
+  - Blacklist: This allows you to order the warnings according to the particular blacklist (i.e. 'Models' or 'Porn' or 'Celebrities') they are associated with. This is helpful if you want to focus in on one particular type of warning at a time.
+  - Keyword: This allows you to order the warnings according to the keywords they are associated with. This is helpful if you want to focus in on warnings for one particular keyword at a time.
+  - Strength: This refers to a number from 0 to 100+, and is the initial value assigned to the warning when NR filtered it the first time, with higher numbers indicating that NR is reporting a greater likelihood of that site being inappropriate. This is helpful if you want to focus on warnings to which NR has assigned either a low strength or a high strength.
+
+!!'Ascending' or 'Descending'
+This option is basically just a way to switch the order of the various 'Order by' options listed above. For options that are based on values or numbers, choosing 'Ascending' will start with the lowest value and move higher. For options that are based on words, 'Ascending' will start with the letter 'A' (or whatever is closest to it) and move through the alphabet to the letter 'Z'. Choosing 'Descending' will do just the opposite of 'Ascending'.
+
+!!'Limit'
+This option allows you to choose how many warnings you want to have loaded at one time. The default is 100 but you may increase it if you want, or decrease it if it's taking too long to load.
+
+!!'Reload'
+If you click on this button, it will reload all warnings. So, if you have simply had warnings appended (either automatically or manually) and you want to start fresh with a new set of warnings, this can be helpful. Or if you have changed any of the options in regards to what types of warnings you want to view and rate, or how you want them sorted, you will need to press this button in order to apply those changes.
+
+!!'Append'
+You can press this button if you want to load more warnings, and have them added to the end of those already displayed. This will happen automatically when you are rating the last few warnings shown.
+
+!!'Show/hide details'
+You can click on this link (or press the key 'd' on your keyboard) to show/hide more details about selected warning(s).
+
+!!'Vote once'
+It is recommended to use this option. If you have it selected, the interface will recognize if you've already rated a warning and avoid displaying it to
+you again in the current or any subsequent rating session. You will need to log in with your NR account in order to choose this option, and also for it to work properly.
+
+
+!!!Warnings display
+You can think of the warnings as being displayed in a table of rows and columns. Each row deals with only one warning, and the various columns give different information about the warning. The following is a short description (from left to right) of the type of information contained in each column.
+
+!!Strength
+In the same way as in the NR reports, the first column contains a color-coded number from 0 to 100+, and is the initial value assigned to the warning when NR filtered it the first time, with higher numbers indicating that NR is reporting a greater likelihood of that site being inappropriate. This value will never be affected by your rating of the warning.
+
+!!Current rating
+The second column contains a number from 0 to 5, representing the current rating of that warning, based on how it has been rated by other users and/or accountability partners. If the number is 0, it means the warning has not yet been rated by anyone. The numbers 1, 2, 3, 4, and 5, represent points along a continuum from something that was definitely a false positive (lower number) to something that was definitely an inappropriate site (higher number). Once a warning has been rated more than once, this number is based on the average rating, so you may see numbers with decimals like 1.23 or 3.45, etc., but they will all be within the 1-5 range if they have been rated before. The rating that you give will be included in calculating a new average rating that will appear in this column. If you choose the 'Show details' option, the number of "votes" will also be displayed within parentheses in this column.
+
+!!Rate
+In the third column, for a warning you have not yet rated, you will see the numbers 1, 2, 3, 4, and 5. You can click on any of these numbers to rate the warning. This can also be done with the keyboard by first [[#17 selecting the warning(s)]] you want to rate and then pressing the number on the keyboard of the rating you want to give. If you choose the 'Show details' option, the blacklist category and keyword for the warning will be displayed in this column
+as well. Once you have rated a warning, only the number you gave it will show up in this column.
+
+!!Abbreviated URL
+The fourth column contains the abbreviated URL with emphasized matches. This is what you will use to determine what rating to give the warning. Hovering over the URL will show the blacklist category and keyword. If you choose the 'Show details' option, the complete URL will display, also with the matches emphasized. Clicking on the URL will not take you to the associated site.
+
+
+!!!How to rate the warnings
+So far, we have looked at the various details and options found on the online interface page for interacting with the improvement data. In this section we will look at how you actually rate the warnings.
+
+!!How do I know what number to rate a warning as?
+We want to see each warning rated according to the following points or markers on a 1-5 continuum:
+  - <!1:!> This is definitely a false positive. The keyword might be present, but it's obvious it wasn't any harm. Example string from URL:
+
+<*cas.be.eu.criteo.com/..._SIMBHNUZBLkvp90fZwLE<!SeX!>vVkp5MPrZms71bFwm8o...*>
+
+  - <!2:!> This was a match, but is unlikely a really dangerous site. Example string from URL:
+
+<*a.tribalfusion.com/...50&addBlockingCategories=<!Adult!>,Political,Flashing...*>
+
+  - <!3:!> You should probably look at the URL to ensure this isn't something to be concerned about. It's a match, but not necessarily dangerous. Example string from URL:
+
+<*upload.wikimedia.org/...150px-<!MariahCarey!>13EdwardsDec1998.jpg*>
+
+  - <!4:!> This is most probably something to worry about. If several of these URLs are listed you should get in touch with the user to check if everything is okay. Example string from URL:
+
+<*www.google.com/...&geo&q="<!sunny!>+<!leone!>+<!sex!>"&gprop&date=today+3-m&...*>
+
+  - <!5:!> This is not only a match, it's <*definitely*> a bad site; you don't even have to think twice, it's really that obvious. Most often a combination of several keywords are found in the same URL. We'd rather not give an example for this level.
+
+<!Note:!> If you have also rated warnings from within an NR report, you will notice that the 1-2-3-4-5 rating continuum detailed above is identical to the 0-25-50-75-100 rating continuum used in the reports, but just using different numbers. The reason for the difference in numbers is because this online interface was developed before the 0-25-50-75-100 levels of strength were set, which are used in the reports. So, even though the numbers are different, the rating is done in the same way. See the [[0705-UnderstandingReports.html#4 'Warnings' section of the "Understanding the Reports" guide]] for more information about the levels of strength used in the reports and how you can rate warnings within a report.
+
+!!Selecting and deselecting warnings
+You can select a warning by clicking on it anywhere except on the numbered links in the third column. When a warning is selected, it will have a grey background. If you want to select multiple warnings that are consecutive, select the first one and then hold down the 'Shift' key while clicking on the last one. If you want to select multiple warnings that are not consecutive, select one of them and then hold down the 'Ctrl' key while clicking on the others. The advantage of selecting multiple warnings is that you can then rate them all at the same time with the same number. If you have multiple warnings selected, you can also deselect any one of them by holding down the 'Ctrl' key while clicking on the one you don't want selected. If you want to deselect all warnings, you can simply click on any warning (without holding any keys down), which will result in only that one being selected.
+
+!!Rating the warnings
+As mentioned above, you can click on the numbered links in the third column in order to rate a warning, but you can also use the numbers on your keyboard to give ratings to whatever warnings are selected at the time. When you press 1, 2, 3, 4, or 5 on the keyboard, all the selected warnings will be given that number rating. Once you do this, the next unrated warning will then be automatically selected. This makes it easier to go through a lot of warnings, simply by selecting one and pressing the keys 1-5 accordingly, or selecting a lot of warnings at the same time and giving them all the same rating. You may also use the up and down arrows on your keyboard to select the next or previous warning.
+
+
+!!!Ideas and tips for avoiding and resisting temptations
+Obviously, if you help with rating this improvement data, you'll likely find a lot of warnings and matches that may pull your thoughts in directions you don't want to go. Because of this, you will want to make sure you do whatever you need to in order to avoid and resist temptation. Here are a couple ideas that might help with this:
+  - Only rate the improvement data when someone else is around. Even better is to tell them explicitly what you're about to do, and ask them to keep an eye on you.
+  - Never do it when you're too tired or are feeling sick. When you're weak physically, it will be easier to fall than at other times.
+  - If you sense that your thoughts are wandering inappropriately, shut down the computer/internet access immediately. It's simply not worth it.
+  - Only rate warnings with a low strength, since these are probably not as harmful. This can be set in the 'Where' field.
+  - When you face warnings that have an obvious rating of 5, don't stop and look at them, just rate and proceed. Refuse to let yourself think about what the image, website, movie etc. looked like in reality. The same thing goes for other rating levels as well. If the level is more or less obvious, then don't take time to investigate it further, just rate and proceed.
+  - Set a time limit on how long you're going to sit down and do this. It might be better to do it in a few shorter intervals than for an extended period of time.
+  - If you for any reason realize that you shouldn't be doing this, then you may blacklist the site by adding it to your personal blacklist.
+

This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.





2012	Jan	Feb	Mar	Apr (24)	May (35)	Jun (5)	Jul (8)	Aug	Sep	Oct (7)	Nov (3)	Dec
2013	Jan	Feb	Mar	Apr	May	Jun	Jul (2)	Aug (1)	Sep	Oct	Nov (1)	Dec
2014	Jan	Feb	Mar	Apr	May (4)	Jun (7)	Jul (4)	Aug	Sep	Oct	Nov	Dec
2015	Jan	Feb (2)	Mar (3)	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec
2016	Jan (1)	Feb (3)	Mar (7)	Apr	May	Jun	Jul	Aug (2)	Sep (4)	Oct (4)	Nov	Dec
2017	Jan	Feb	Mar (2)	Apr (2)	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec (1)
2018	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep (1)	Oct	Nov	Dec
2019	Jan	Feb	Mar	Apr (2)	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec

responsibility-commit Mailing List for Net Responsibility (Page 4)

Free Accountability Software

responsibility-commit — Subversion commit notification