SUMO community discussions
New Feature Proposal: Mark Thread as Duplicated
-
Hi everyone!
Something I've been thinking about is to include the possibility of marking a thread in the Support forum as a duplicate of another thread. The main goals for this feature are three:
- Simplify and speed up the delivery of an answer.
By marking as duplicate the user who ask a question could receive the answer from the other thread, while it would be extra simple for a contributor to point to that other thread (i.e. simply include the Thread ID in a box or similar).
- Create a new way start contributing to SUMO.
Users who want to start to help other users may be a little bit overwhelm with the actual system. By allowing to contribute by including links to other threads we are technically reducing the entry bar.
- Surface threads that are commonly asked.
At the moment, the only way to identify threads that are popular is by the amount of "Me too's". If we can use duplicates as a symptom of relevancy we could improve the results that we are showing in our search results. Even better, we could avoid showing "duplicated" threads and only show one of the versions (the one that is solved). This is just one example.
What do you think? Is this something worth investigating? We should land the design of the feature if we think is worth implementing.
Hi everyone!
Something I've been thinking about is to include the possibility of marking a thread in the Support forum as a duplicate of another thread. The main goals for this feature are three:
* Simplify and speed up the delivery of an answer.
By marking as duplicate the user who ask a question could receive the answer from the other thread, while it would be extra simple for a contributor to point to that other thread (i.e. simply include the Thread ID in a box or similar).
* Create a new way start contributing to SUMO.
Users who want to start to help other users may be a little bit overwhelm with the actual system. By allowing to contribute by including links to other threads we are technically reducing the entry bar.
* Surface threads that are commonly asked.
At the moment, the only way to identify threads that are popular is by the amount of "Me too's". If we can use duplicates as a symptom of relevancy we could improve the results that we are showing in our search results. Even better, we could avoid showing "duplicated" threads and only show one of the versions (the one that is solved). This is just one example.
What do you think? Is this something worth investigating? We should land the design of the feature if we think is worth implementing.
-
I think this would be cool. We could reward karma points for finding duplicates. Also, it would be nice if we could use search to identify probable questions that it is a dupe of.
I think this would be cool. We could reward karma points for finding duplicates. Also, it would be nice if we could use search to identify probable questions that it is a dupe of.
-
I've been thinking about is to include the possibility of marking a thread in the Support forum as a duplicate of another thread.
Is the original process of entering a question already supposed to flag up such posts ?
- it may not always
is there any way of including some user input ?
as an example could we ask if a suggested answer is relevant and score that resulting input for comparison with the scoring of the algorithm that made the original suggestion
- if the only person seeing those suggestions is the user asking a question
the quality of the suggestion can only be compared to the ability to provide a correct answer. IF on the other hand those suggestions were recorded & flagged alongside the question as links others could rate them. Most suggestions would likely be useless probably generic over simplistic answers not addressing the specifics. Maybe you could get usable info from suggestions, if they could somehow be scored.
That however produces a problem similar to the helpful scores often being meaningless
- high scores often mean a rant rather than a useful answer
- (I surmise) a very clever algorithm may differentiate; but opinion from trusted contributors, or the rarely available mods, is likely to give a much more accurate result than any initial algorithm.
So for concrete suggestions about sumo
- record and flag the suggestions provided when the user asks a question
- have a method for rating those suggestions - rather than continuing to blindly provide info of dubious utility
I am not going to start bug spam, but what about considering something for the low hanging fruit bug discussion?
<blockquote>I've been thinking about is to include the possibility of marking a thread in the Support forum as a duplicate of another thread. </blockquote>
Is the original process of entering a question already supposed to flag up such posts ?
* it may not always <br/>'''is there any way of including some user input ? '''<br/> as an example could we ask if a suggested answer is relevant and score that resulting input for comparison with the scoring of the algorithm that made the original suggestion
* if the only person seeing those suggestions is the user asking a question <br/>the quality of the suggestion can only be compared to the ability to provide a correct answer. IF on the other hand those suggestions were recorded & flagged alongside the question as links others could rate them. Most suggestions would likely be useless probably generic over simplistic answers not addressing the specifics. Maybe you could get usable info from suggestions, if they could somehow be scored.
That however produces a problem similar to the helpful scores often being meaningless
* high scores often mean a rant rather than a useful answer
* (I surmise) a very clever algorithm may differentiate; but opinion from trusted contributors, or the rarely available mods, is likely to give a much more accurate result than any initial algorithm.
''' So for concrete suggestions about sumo'''
* record and flag the suggestions provided when the user asks a question
* have a method for rating those suggestions - rather than continuing to blindly provide info of dubious utility
I am not going to start bug spam, but what about considering something for the low hanging fruit bug discussion?
-
Not a good idea in my opinion. Mods or contributors may provide links, the average user is (I presume: you have the metrics) a one off question, there are too many variables. All it will achieve is the rants getting linked together , not the useful answers.
Rants of
- Fx #X is xxxx
- High Memory use
- Something does not work
are all very easy to link together, but often many variables need to be considered, if someone thinks carefully and makes a reasoned link fine, but do not encourage indiscriminate linking.
Your idea could work, but not in the current forum where most questions are either unanswered or unsolved. If anything in my opinion the two primary goals should be
- prevent duplicated and unnecessary questions
- recruit more volunteers
Not a good idea in my opinion. Mods or contributors may provide links, the average user is (I presume: you have the metrics) a one off question, there are too many variables. All it will achieve is the rants getting linked together , not the useful answers.
Rants of
*Fx #X is xxxx
* High Memory use
* Something does not work
are all very easy to link together, but often many variables need to be considered, if someone thinks carefully and makes a reasoned link fine, but do not encourage indiscriminate linking.
Your idea could work, but not in the current forum where most questions are either unanswered or unsolved. If anything in my opinion the two primary goals should be
* prevent duplicated and unnecessary questions
* recruit more volunteers
-
John99 said
prevent duplicated and unnecessary questions
To do that reviewers must approve articles in pending review at least first the most unhelpful.
''John99 [[#post-42660|said]]''
<blockquote>
prevent duplicated and unnecessary questions
</blockquote>
To do that reviewers must approve [/contributors/unreviewed articles in pending review] at least first [/contributors/unhelpful the most unhelpful].
-
John99 said
Not a good idea in my opinion. Mods or contributors may provide links, the average user is (I presume: you have the metrics) a one off question, there are too many variables. All it will achieve is the rants getting linked together , not the useful answers.
Rants of
- Fx #X is xxxx
- High Memory use
- Something does not work
are all very easy to link together, but often many variables need to be considered, if someone thinks carefully and makes a reasoned link fine, but do not encourage indiscriminate linking.
Two thing here. As I said, you are assuming that in the flow none reviews duplicates and anyone has the ability to do it. I think that we can solve this really easily by confirming duplicates by contributors and after X amount of successful duplicates you get that ability too.
I don't expect regular users to mark duplicates but is an option for people who wants to contribute and it's yet not super confident about giving the right answer.
The second point is that contributors can give links already. That's true. What we are trying to achieve here is to standardize that and give extra value to those provided links. At the moment if you include a link to another thread as an answer, it's just a link to another thread. With this feature it will be a link to the other thread plus a signal that we can use to surface threads.
John99 said
Your idea could work, but not in the current forum where most questions are either unanswered or unsolved. If anything in my opinion the two primary goals should be
- prevent duplicated and unnecessary questions
- recruit more volunteers
And from my experience this is (at least partially) because the amount of duplicates. I'm more keen to answer new questions than duplicated ones and those that are already well documented.
''John99 [[#post-42660|said]]''
<blockquote>
Not a good idea in my opinion. Mods or contributors may provide links, the average user is (I presume: you have the metrics) a one off question, there are too many variables. All it will achieve is the rants getting linked together , not the useful answers.
Rants of
*Fx #X is xxxx
* High Memory use
* Something does not work
are all very easy to link together, but often many variables need to be considered, if someone thinks carefully and makes a reasoned link fine, but do not encourage indiscriminate linking.
</blockquote>
Two thing here. As I said, you are assuming that in the flow none reviews duplicates and anyone has the ability to do it. I think that we can solve this really easily by confirming duplicates by contributors and after X amount of successful duplicates you get that ability too.
I don't expect regular users to mark duplicates but is an option for people who wants to contribute and it's yet not super confident about giving the right answer.
The second point is that contributors can give links already. That's true. What we are trying to achieve here is to standardize that and give extra value to those provided links. At the moment if you include a link to another thread as an answer, it's just a link to another thread. With this feature it will be a link to the other thread plus a signal that we can use to surface threads.
''John99 [[#post-42660|said]]''
<blockquote>
Your idea could work, but not in the current forum where most questions are either unanswered or unsolved. If anything in my opinion the two primary goals should be
* prevent duplicated and unnecessary questions
* recruit more volunteers
</blockquote>
And from my experience this is (at least partially) because the amount of duplicates. I'm more keen to answer new questions than duplicated ones and those that are already well documented.
-
Ibai said
And from my experience this is (at least partially) because the amount of duplicates.
Did you count the amount of real duplicates over one day (about 400 questions)? 5%, 25%, 50%? By real duplicates, I mean it's not because two questions have similar topics that the solution is the same and they are therefore duplicated. For instance with two questions with "Firefox crashes" as title, a solution for one would be to disable a crashy extension and for the second one to update graphic drivers.
From my point of view (I preach again for my KB editor church), duplicates are a consequence of KB articles with missing solutions because not updated regularly. The KB update must be done with two sources:
- using support forum contributor feedbacks;
- looking for untreated solutions in support forums.
With a KB close to the problems users have, you won't prevent users from asking a question even if the answer is in the KB, but the amount of unnecessary questions will be at least lower than today.
''Ibai [[#post-42668|said]]''
<blockquote>
And from my experience this is (at least partially) because the amount of duplicates.
</blockquote>
Did you count the amount of real duplicates over one day (about 400 questions)? 5%, 25%, 50%? By real duplicates, I mean it's not because two questions have similar topics that the solution is the same and they are therefore duplicated. For instance with two questions with "Firefox crashes" as title, a solution for one would be to disable a crashy extension and for the second one to update graphic drivers.
From my point of view (I preach again for my KB editor church), duplicates are a consequence of KB articles with missing solutions because not updated regularly. The KB update must be done with two sources:
* using support forum contributor feedbacks;
* looking for untreated solutions in support forums.
With a KB close to the problems users have, you won't prevent users from asking a question even if the answer is in the KB, but the amount of unnecessary questions will be at least lower than today.
-
I haven't count them in the general forum. It's complicated because you can't only measure duplicates of today against today but today against the whole history.
Your point is completely valid, and it's one of the focus we should keep having. That said, there's a good amount of users that whatever we do, they will want to ask their question and don't want to read generalist content. And there's another thing that we can't tackle 100% with KB articles is users who don't know how to phrase their issues. So, IMHO, they are solutions that we can combine for the best outcome.
I haven't count them in the general forum. It's complicated because you can't only measure duplicates of today against today but today against the whole history.
Your point is completely valid, and it's one of the focus we should keep having. That said, there's a good amount of users that whatever we do, they will want to ask their question and don't want to read generalist content. And there's another thing that we can't tackle 100% with KB articles is users who don't know how to phrase their issues. So, IMHO, they are solutions that we can combine for the best outcome.
-
Ibaii,
May I ask who is going to mark the threads as a duplicate under your proposal ?
How are these to be found ?
Who is going to do the finding ?
yes it could be useful, but it could not work or be open to abuse.
The usual and probably acknowledged problem are that
- often a user will not look at possible answers but instead ask a question
- most suggested answers are not appropriate
IF kb articles are up to date then surely the aim is they answer the majority of questions.
I doubt linking support question answers will ever answer the majority of questions, BUT I also doubt they KB s will ever answer immediately some detailed questions. IMHO we need a two pronged approach
- keep KB articles up-to-date
- rapid update put out answers quickly, even if only in stub form, as new problems emerge, eg as a new release is made
- provide provisional answers make users aware there are sometimes discussions and revisions pending besides the fully approved answers, be more open instead of striving for perfection
- improve sumo and how articles are rated
- not entirely sure what to suggest the current system has major failings
- many questions never get answered
- most users never give feedback
- back to your original problem, many questions have an answer and have been asked before
Ibaii,
May I ask who is going to mark the threads as a duplicate under your proposal ?
How are these to be found ?
Who is going to do the finding ?
yes it could be useful, but it could not work or be open to abuse.
The usual and probably acknowledged problem are that
*often a user will not look at possible answers but instead ask a question
* most suggested answers are not appropriate
IF kb articles are up to date then surely the aim is they answer the majority of questions.
I doubt linking support question answers will ever answer the majority of questions, BUT I also doubt they KB s will ever answer immediately some detailed questions. IMHO we need a two pronged approach
# keep KB articles up-to-date
#* <u> rapid update</u> put out answers quickly, even if only in stub form, as new problems emerge, eg as a new release is made
#* <u>provide provisional answers</u> make users aware there are sometimes discussions and revisions pending besides the fully approved answers, be more open instead of striving for perfection
# improve sumo and how articles are rated
#* not entirely sure what to suggest the current system has major failings
** many questions never get answered
** most users never give feedback
** back to your original problem, many questions have an answer and have been asked before
-
This will be controversial, and is not necessarily well thought out.
How about imposing some cost on asking a question ? with the aim of reducing the number of unnecessary questions ? Have a two star premium or one star standard service, the user chooses
Premium
- monetary - have a premium channel someone who has made a donation and can ask a question even if it is incomplete.
- full information - make an input form - mandatory fields have to be completed
- knowledgeable - able to confirm have read other articles or questions
Yes easy to cheat, but if we ask someone to read say the top three answers, they may find it works and then they do not need to ask a question, or will ask a more informed question. We could enforce a requirent to click on the articles, much like EULA's do.
- reward contributors - including potential contributors
dangle the promise that the user gets premium rate questions if they provide feedback on the original question
Standard
- as now - most users likely to post incomplete information and not look first at other answers
It is probably what happens already, someone asking a poorly formatted incomplete question is less likely to get any reply at all. Why not just make that clear by saying there are two services.
This will be controversial, and is not necessarily well thought out.
How about imposing some cost on asking a question ? with the aim of reducing the number of unnecessary questions ? Have a two star premium or one star standard service, the user chooses
Premium
* monetary - have a premium channel someone who has made a donation and can ask a question even if it is incomplete.
* full information - make an input form - mandatory fields have to be completed
* knowledgeable - able to confirm have read other articles or questions <br/>Yes easy to cheat, but if we ask someone to read say the top three answers, they may find it works and then they do not need to ask a question, or will ask a more informed question. We could enforce a requirent to click on the articles, much like EULA's do.
* reward contributors - including potential contributors <br/>
dangle the promise that the user gets premium rate questions if they provide feedback on the original question
Standard
- as now - most users likely to post incomplete information and not look first at other answers
It is probably what happens already, someone asking a poorly formatted incomplete question is less likely to get any reply at all. Why not just make that clear by saying there are two services.
-
As far as true duplicate postings where a users asks the same question two or more times on the same day (and creating multiple threads), usually withing minutes of each other, the forum software need to be fixed! Users should not be expected to reload the "answers" main page repeatedly to see the thread that they just started. I have seen a lag happen where a full 5 minutes after answering a thread it still doesn't appear at the top of the first "answers" page. That is probably why we are seeing so many duplicate posting by the same new user of this forum. And it seems to run in cycles at certain times of the day on many days. Overall, most days it seems that it happens at least a dozen times or more that I have seen and marked as duplicates.
mha007 and SafeBrowser have been PM'ing me with a hyperlink of the thread that needs to be locked as a duplicate when they have been finding duplicate thread and marking them as such, so I can lock them when I come onboard.
As far as true duplicate postings where a users asks the same question two or more times on the same day (and creating multiple threads), usually withing minutes of each other, the forum software need to be fixed! Users should not be expected to reload the "answers" main page repeatedly to see the thread that they just started. I have seen a lag happen where a full 5 minutes after answering a thread it still doesn't appear at the top of the first "answers" page. That is probably why we are seeing so many duplicate posting by the same new user of this forum. And it seems to run in cycles at certain times of the day on many days. Overall, most days it seems that it happens at least a dozen times or more that I have seen and marked as duplicates.
'''mha007''' and '''SafeBrowser''' have been PM'ing me with a hyperlink of the thread that needs to be locked as a duplicate when they have been finding duplicate thread and marking them as such, so I can lock them when I come onboard.
-
Good point edmeister, but possibly a minor cause of duplicates. If I come across such duplicates. I do try to link them together, so that only one gets answered, if it helps I will also pm you when I come across these in future.
Good point edmeister, but possibly a minor cause of duplicates. If I come across such duplicates. I do try to link them together, so that only one gets answered, if it helps I will also pm you when I come across these in future.
-
ATTN admin: Gentle reminder to review the most unhelpful articles in pending review.
'''ATTN admin:''' Gentle reminder to review the most [/contributors/unhelpful unhelpful articles in pending review].
