word2vec generating results!

[peparedes]

Guys, good news!!

@lambdaloop and @xih managed to merge their matrices and magic is being done. @xih created a matrix of words (taken from the master dict) to vectors (taken from the Google word2vec corpus). @lambdaloop created a bag of sentences mapped onto words (a very sparse one of course). The fusion (multiplication) of both created a matrix of sentences mapped into vectors.

Our first query with the word "cancer" generated very interesting sentences, which seem to be adequate to be tagged as new seeds for more precise life events. We need to now find a way to preserve the order of the words within a sentence. Correct some punctuation issues, and record the original userIDs and postIDs to further analyze the outcomes and tag them as valid or not. @lambdaloop will look into this and will reach out to @coryschillaci and @anasrferreira to find the best way to retrieve that info.

We will need to find a good way to do the initial manual tagging. @lambdaloop will create a simple list of some outcomes to start tagging them. One option we are thinking is what if we can give web access to our lucene implementation so we can post the links of the selected postIDs to the web to get MTurkers to tag the sentences for us. We could even ask MTurkers to tag them not only in a binary form, but even rate the level of "feeling" each phrase has. @davclark and @geneyoo could you look into this maybe?

[coryschillaci]

@lambdaloop Are you happy with the sentence by sentence featurizer?

[lambdaloop]

@coryschillaci Do you see something wrong with it?

One problem with it is that it maps the punctation from the previous sentence, instead of the current one. To fix that, I would need to shift the sentence indices by one, but I'm not sure how to do that yet in Scala (it's probably trivial).

Another problem which Pablo touched on above is that I also want to store the word order of the sentences somehow, so that we can see later which sentences got matched.

[lambdaloop]

By the way, the sentence by sentence featurizer code is in here: https://github.com/berkeley-dsc/destress/blob/master/process_data/featurizers.scala

The rest of the word2vec effort (@xih's code to generate a matrix word2vec vectors for the words in our master dictionary and an initial query script) is in the word2vec directory: https://github.com/berkeley-dsc/destress/tree/master/word2vec

[peparedes]

@lambdaloop, could you please send a histogram of the sentence lengths.

Thanks,

Pablo On Apr 16, 2015 3:19 PM, "Pierre Karashchuk" notifications@github.com wrote:

@coryschillaci https://github.com/coryschillaci Do you see something wrong with it?

One problem with it is that it maps the punctation from the previous sentence, instead of the current one. To fix that, I would need to shift the sentence indices by one, but I'm not sure how to do that yet in Scala (it's probably trivial).

Another problem which Pablo touched on above is that I also want to store the word order of the sentences somehow, so that we can see later which sentences got matched.

— Reply to this email directly or view it on GitHub https://github.com/berkeley-dsc/destress/issues/33#issuecomment-93852252 .

[lambdaloop]

@peparedes Oh, I haven't run the featurizer on the whole dataset yet. I'll run it tonight and report.

[geneyoo]

@peparedes I recall reading about Solr (which is very similar to Lucene, but is a java web-app version). This may be what we're looking for, to give MTurkers access to the full-text search, though I am not sure because our files& indices would still be on a private server. If permissions are not an issue, I will be happy to set up another full-text search via Solr.

@davclark do you have any info on this?

[lambdaloop]

It works!!!

I've made a query script, which we can use to match a word or sentence to the database using word2vec.

You can call it like this (on mercury):

bash /home/pierre/query.sh

Then you will be put onto the scala shell with BIDMach bindings and a query function. query takes 2 arguments, the word(s) to match and the number of top entries to return.

Examples:

query("cancer", 10)  // top 10 matches for cancer
query("my dad died", 20) // top 20 matches for (my + dad + died)

[peparedes]

Awesome!! Will try it asap.

P

On Thu, Apr 16, 2015 at 11:34 PM, Pierre Karashchuk < notifications@github.com> wrote:

It works!!!

I've made a query script, which we can use to match a word or sentence to the database using word2vec.

You can call it like this (on mercury):

bash /home/pierre/query.sh

Then you will be put onto the scala shell with BIDMach bindings and a query function. query takes 2 arguments, the word(s) to match and the number of top entries to return.

Examples:

query("cancer", 10) // top 10 matches for cancer query("my dad died", 20) // top 20 matches for (my + dad + died)

— Reply to this email directly or view it on GitHub https://github.com/berkeley-dsc/destress/issues/33#issuecomment-93909494 .

[lambdaloop]

Some initial results:

scala> query("stress", 10)
adding stress to vector

0.835 -- ... because of the stress . 
0.810 -- no stress . 
0.788 -- stress city ! 
0.766 -- aaaaaaahhhhhhhhh the pain and suffering and stress ! 
0.743 -- there s also the fact that i need to start saving money ... stress , stress stress ... 
0.724 -- too much stress . 
0.721 -- my stress and anxiety was growing . 
0.718 -- i have stress migranes ! 
0.715 -- what lowers your stress blood pressure anxiety level ? 
0.715 -- what lowers your stress blood pressure anxiety level ? 

res4: Int = 10

scala> query("stressed", 10)
adding stressed to vector

0.872 -- so stressed . 
0.816 -- stressed much ! 
0.657 -- stressed to the max . 
0.656 -- said . 
0.656 -- said . 
0.580 -- durkin said that . 
0.553 -- he said . 
0.553 -- he said . 
0.545 -- he stated . 
0.528 -- i am always stressed out and worried . 

res5: Int = 10

scala> query("cancer", 10)
adding cancer to vector

0.818 -- cancer : just one . 
0.810 -- its for brest cancer . 
0.787 -- but dorcas , has cancer ? 
0.704 -- he has cancer and going for chemotherapy soon . 
0.674 -- my father was diagnosed with cancer in early june lung cancer that has spread to the bones . 
0.654 -- :( what if it s breast cancer ? 
0.648 -- one out of eight women are diagnosed with breast cancer . 
0.647 -- she had jaw cancer . 
0.637 -- by contrast , were from diseases of the heart , were from malignant neoplasms cancer , and from cerebrovascular disease stroke , the three leading causes . 
0.634 -- from the cancer , i think . 

res6: Int = 10

scala> query("breast cancer", 10)
adding breast to vector
adding cancer to vector

0.794 -- :( what if it s breast cancer ? 
0.740 -- one out of eight women are diagnosed with breast cancer . 
0.731 -- breast implants ? 
0.682 -- cancer : just one . 
0.673 -- its for brest cancer . 
0.654 -- she had jaw cancer . 
0.644 -- but dorcas , has cancer ? 
0.619 -- i borrowed a model of a breast with lumps in it , to demonstrate breast self examination . 
0.611 -- my father was diagnosed with cancer in early june lung cancer that has spread to the bones . 
0.607 -- their dog has a medical problem and they found out that aunt peggy has breast cancer . 

res7: Int = 10

scala> query("sadness", 10)
adding sadness to vector

0.856 -- sorrow ... just sorrow . 
0.807 -- sorrow sing sorrow . 
0.778 -- in silent sadness . 
0.774 -- sorrow ... sweet intoxicating sorrow ! 
0.774 -- the sadness continues . 
0.773 -- they were tears of joy and sadness . 
0.756 -- obligation to sadness ? 
0.738 -- sorrow , sweet sweet sorrow . 
0.737 -- full of joy , sadness , and insanity . 
0.729 -- i felt excitement , sadness , revulsion , fear , happiness , jubilation even . 

res9: Int = 10

scala> query("died", 10)
adding died to vector

0.880 -- and almost died . 
0.879 -- my car died died . 
0.826 -- arthur schlesinger died . 
0.803 -- baby died . 
0.790 -- students died , teachers died . 
0.787 -- i died . 
0.782 -- he died alone . 
0.770 -- she almost died . 
0.770 -- he died a painful death . 
0.757 -- hope he died . 

res10: Int = 10

scala> query("father died", 10)
adding father to vector
adding died to vector

0.876 -- and died . 
0.837 -- my car died died . 
0.809 -- baby died . 
0.791 -- and almost died . 
0.772 -- he died alone . 
0.770 -- he died a painful death . 
0.753 -- she almost died . 
0.749 -- killed his wife ? 
0.749 -- killed his wife ? 
0.742 -- my dad almost died today . 

res11: Int = 10

scala> query("my father died", 10)
adding my to vector
adding father to vector
adding died to vector

0.894 -- my dad almost died today . 
0.877 -- my grandfather on my mother s side has died . 
0.874 -- my car died died . 
0.842 -- : my father . 
0.842 -- my father . 
0.840 -- today , a year ago my great uncle died . 
0.840 -- and apparently , my boss died . 
0.825 -- my computer died . 
0.819 -- one of my dad s friends died thursday . 
0.813 -- in september of it was my grandmother , my great uncle , and my uncle , who all died within the same weeks . 

res12: Int = 10

[lambdaloop]

The first number is the correlation of the search vector with the sentence.

[peparedes]

Very cool!... Try some sentences derived from the LEQ! P

On Thu, Apr 16, 2015 at 11:36 PM, Pierre Karashchuk < notifications@github.com> wrote:

Some initial results:

scala> query("stress", 10) adding stress to vector

0.835 -- ... because of the stress . 0.810 -- no stress . 0.788 -- stress city ! 0.766 -- aaaaaaahhhhhhhhh the pain and suffering and stress ! 0.743 -- there s also the fact that i need to start saving money ... stress , stress stress ... 0.724 -- too much stress . 0.721 -- my stress and anxiety was growing . 0.718 -- i have stress migranes ! 0.715 -- what lowers your stress blood pressure anxiety level ? 0.715 -- what lowers your stress blood pressure anxiety level ?

res4: Int = 10

scala> query("stressed", 10) adding stressed to vector

0.872 -- so stressed . 0.816 -- stressed much ! 0.657 -- stressed to the max . 0.656 -- said . 0.656 -- said . 0.580 -- durkin said that . 0.553 -- he said . 0.553 -- he said . 0.545 -- he stated . 0.528 -- i am always stressed out and worried .

res5: Int = 10

scala> query("cancer", 10) adding cancer to vector

0.818 -- cancer : just one . 0.810 -- its for brest cancer . 0.787 -- but dorcas , has cancer ? 0.704 -- he has cancer and going for chemotherapy soon . 0.674 -- my father was diagnosed with cancer in early june lung cancer that has spread to the bones . 0.654 -- :( what if it s breast cancer ? 0.648 -- one out of eight women are diagnosed with breast cancer . 0.647 -- she had jaw cancer . 0.637 -- by contrast , were from diseases of the heart , were from malignant neoplasms cancer , and from cerebrovascular disease stroke , the three leading causes . 0.634 -- from the cancer , i think .

res6: Int = 10

scala> query("breast cancer", 10) adding breast to vector adding cancer to vector

0.794 -- :( what if it s breast cancer ? 0.740 -- one out of eight women are diagnosed with breast cancer . 0.731 -- breast implants ? 0.682 -- cancer : just one . 0.673 -- its for brest cancer . 0.654 -- she had jaw cancer . 0.644 -- but dorcas , has cancer ? 0.619 -- i borrowed a model of a breast with lumps in it , to demonstrate breast self examination . 0.611 -- my father was diagnosed with cancer in early june lung cancer that has spread to the bones . 0.607 -- their dog has a medical problem and they found out that aunt peggy has breast cancer .

res7: Int = 10

scala> query("sadness", 10) adding sadness to vector

0.856 -- sorrow ... just sorrow . 0.807 -- sorrow sing sorrow . 0.778 -- in silent sadness . 0.774 -- sorrow ... sweet intoxicating sorrow ! 0.774 -- the sadness continues . 0.773 -- they were tears of joy and sadness . 0.756 -- obligation to sadness ? 0.738 -- sorrow , sweet sweet sorrow . 0.737 -- full of joy , sadness , and insanity . 0.729 -- i felt excitement , sadness , revulsion , fear , happiness , jubilation even .

res9: Int = 10

scala> query("died", 10) adding died to vector

0.880 -- and almost died . 0.879 -- my car died died . 0.826 -- arthur schlesinger died . 0.803 -- baby died . 0.790 -- students died , teachers died . 0.787 -- i died . 0.782 -- he died alone . 0.770 -- she almost died . 0.770 -- he died a painful death . 0.757 -- hope he died .

res10: Int = 10

scala> query("father died", 10) adding father to vector adding died to vector

0.876 -- and died . 0.837 -- my car died died . 0.809 -- baby died . 0.791 -- and almost died . 0.772 -- he died alone . 0.770 -- he died a painful death . 0.753 -- she almost died . 0.749 -- killed his wife ? 0.749 -- killed his wife ? 0.742 -- my dad almost died today .

res11: Int = 10

scala> query("my father died", 10) adding my to vector adding father to vector adding died to vector

0.894 -- my dad almost died today . 0.877 -- my grandfather on my mother s side has died . 0.874 -- my car died died . 0.842 -- : my father . 0.842 -- my father . 0.840 -- today , a year ago my great uncle died . 0.840 -- and apparently , my boss died . 0.825 -- my computer died . 0.819 -- one of my dad s friends died thursday . 0.813 -- in september of it was my grandmother , my great uncle , and my uncle , who all died within the same weeks .

res12: Int = 10

— Reply to this email directly or view it on GitHub https://github.com/berkeley-dsc/destress/issues/33#issuecomment-93910205 .

[lambdaloop]

scala> query("have drinking problem", 10);
adding have to vector
adding drinking to vector
adding problem to vector

0.745 -- drinking . 
0.726 -- he needs to stop drinking though . 
0.723 -- no drinking though . 
0.719 -- without drinking . 
0.718 -- so much for drinking . 
0.715 -- too drunk and , still drinking ! 
0.715 -- but i have done the whole drinking alcohol thing . 
0.713 -- and he laufed and then he s like so do u have a drinking problem ? 
0.711 -- if he has such a problem with me drinking ... he shouldn t smoke . 
0.704 -- lots of drinking . 

scala> query("I feel depressed", 10)
adding i to vector
adding feel to vector
adding depressed to vector

1.000 -- i feel depressed . 
0.870 -- i feel really depressed right now . 
0.863 -- why do i feel so depressed ? 
0.859 -- i think im depressed . 
0.855 -- i feel corageous . 
0.854 -- i m not even feeling depressed lately . 
0.853 -- i am depressed again i think . 
0.850 -- i feel happy i feel depressed i feel angry and mad i feel love i feel hate i feel pain ... all at once ... i don t know whether to tell how i feel or to keep it all inside . 
0.847 -- i m so tired of being depressed , feeling like i dont matter . 
0.844 -- i feel lonely . 

scala> query("I slept like a baby.", 10)
adding i to vector
adding slept to vector
adding like to vector
WARNING: a is not in google wordvec database
WARNING: did not find baby. in master dict

0.934 -- i slept . 
0.919 -- and then ... i slept . 
0.915 -- i slept in . 
0.912 -- i slept good . 
0.893 -- i slept a lot . 
0.889 -- i slept until like : and then i ate . 
0.867 -- i slept really good last night , but i feel like i haven t slept at all . 
0.867 -- but when i got home i slept like anythings . 
0.858 -- i guess i slept it off . 
0.857 -- i slept in today . 

scala> query("I have mild insomnia", 10)
adding i to vector
adding have to vector
adding mild to vector
adding insomnia to vector

0.826 -- i have mild moderate depression . 
0.781 -- ok i think i have insomnia . 
0.775 -- but i might be suffering from temporary insomnia . 
0.771 -- unfortunate to admit , but it s true ... well ... insomnia or go to bed ... i choose insomnia ! 
0.745 -- insomnia , insomnia must sleep on it to think it over ... 
0.742 -- i also think that s why i have insomnia . 
0.738 -- i don t have insomnia . 
0.733 -- im tired ... i think i have sleep insomnia or something . 
0.705 -- i m probably overreacting , but i can t control my insomnia . 
0.703 -- i think i may have narcolepsy ... oh well . 

scala> query("horrible abortion", 10)
adding horrible to vector
adding abortion to vector

0.863 -- abortion ? 
0.827 -- yes abortion ? 
0.743 -- had an abortion . 
0.730 -- they like abortions . 
0.698 -- abortion : for or against it ? 
0.657 -- i dont really believe in abortion . 
0.628 -- a horrible horrible death . 
0.620 -- no what do you think about abortion ? 
0.619 -- abortion is wrong : in some cases , in others , no . 
0.617 -- horrible ? 

scala> query("gross fuckin miscarriage", 10)
adding gross to vector
adding fuckin to vector
adding miscarriage to vector

0.705 -- it was fucking gross . 
0.658 -- : nope ... gross 
0.649 -- and hayleys like oh gross ! 
0.639 -- fuckin . 
0.638 -- so gross . 
0.632 -- gross , i know . 
0.631 -- gross , i d fucking kill him . 
0.629 -- fuckin bitch . 
0.628 -- shes fucking gross now with her blonde hair and her gross skinny ness . 
0.626 -- gross . 

scala> query("Nobody will hire me", 10)
adding nobody to vector
adding will to vector
adding hire to vector
adding me to vector

0.843 -- the other bad thing is people will still hire me , not too many will chance hiring my son 
0.838 -- walgreens might hire me . 
0.826 -- i really hope they hire me . 
0.823 -- if anyone wants to go with me let me know . 
0.819 -- if anyone wants to come , let me know ! 
0.818 -- will someone help me ? 
0.818 -- i also wanted to ask if they had any openings to hire me . 
0.818 -- if anybody wants to do something give me a call ! 
0.813 -- but i will always have someone who wants me . 
0.812 -- maybe god will present him to me again someday and tell me that he is right for me . 

scala> query("READY FOR SCHOOL", 10)
adding ready to vector
adding for to vector
adding school to vector

0.889 -- getting ready for school and all . 
0.808 -- horray for school . 
0.805 -- im not ready for school . 
0.804 -- i got ready for school . 
0.791 -- i should get ready for school now . 
0.770 -- oh well ... time to get ready for school tommorow ! 
0.755 -- i have to get ready for school now , though . 
0.747 -- today after school all my friends are getting ready for homecoming . 
0.734 -- school for another week . 
0.729 -- alright , well time for school . 

scala> query("I like my new neighbors", 10)
adding i to vector
adding like to vector
adding my to vector
adding new to vector
adding neighbors to vector

0.891 -- i like my friends . 
0.873 -- i just like my jmu . 
0.867 -- i really like my new things . 
0.867 -- i like my friends and kelsey ahha . 
0.859 -- i do it with my friends . 
0.858 -- i feel like i have failed my friends for the month that i wasnt me . 
0.858 -- i like seeing my friends though . 
0.858 -- i like my sister 
0.857 -- i love my neighbors , seriously . 
0.857 -- i love it ... i like my swiffer wetjet too yes , i have my own , shut up . 

scala> query("cheated", 10)
adding cheated to vector

0.782 -- been cheated on ? 
0.747 -- cheated on someone ? 
0.745 -- cheated in a relationship . 
0.726 -- : idk ever cheated ? 
0.722 -- : yes been cheated on ? 
0.698 -- : yep been cheated on ? 
0.693 -- yup been cheated on ? 
0.689 -- ever been cheated on ? 
0.682 -- cheated on a test guilty . 
0.681 -- yeah been cheated on ? 

[peparedes]

very interesting! Some have pretty good answers!

P

On Fri, Apr 17, 2015 at 12:23 AM, Pierre Karashchuk < notifications@github.com> wrote:

scala> query("have drinking problem", 10); adding have to vector adding drinking to vector adding problem to vector

0.745 -- drinking . 0.726 -- he needs to stop drinking though . 0.723 -- no drinking though . 0.719 -- without drinking . 0.718 -- so much for drinking . 0.715 -- too drunk and , still drinking ! 0.715 -- but i have done the whole drinking alcohol thing . 0.713 -- and he laufed and then he s like so do u have a drinking problem ? 0.711 -- if he has such a problem with me drinking ... he shouldn t smoke . 0.704 -- lots of drinking .

scala> query("I feel depressed", 10) adding i to vector adding feel to vector adding depressed to vector

1.000 -- i feel depressed . 0.870 -- i feel really depressed right now . 0.863 -- why do i feel so depressed ? 0.859 -- i think im depressed . 0.855 -- i feel corageous . 0.854 -- i m not even feeling depressed lately . 0.853 -- i am depressed again i think . 0.850 -- i feel happy i feel depressed i feel angry and mad i feel love i feel hate i feel pain ... all at once ... i don t know whether to tell how i feel or to keep it all inside . 0.847 -- i m so tired of being depressed , feeling like i dont matter . 0.844 -- i feel lonely .

scala> query("I slept like a baby.", 10) adding i to vector adding slept to vector adding like to vector WARNING: a is not in google wordvec database WARNING: did not find baby. in master dict

0.934 -- i slept . 0.919 -- and then ... i slept . 0.915 -- i slept in . 0.912 -- i slept good . 0.893 -- i slept a lot . 0.889 -- i slept until like : and then i ate . 0.867 -- i slept really good last night , but i feel like i haven t slept at all . 0.867 -- but when i got home i slept like anythings . 0.858 -- i guess i slept it off . 0.857 -- i slept in today .

scala> query("I have mild insomnia", 10) adding i to vector adding have to vector adding mild to vector adding insomnia to vector

0.826 -- i have mild moderate depression . 0.781 -- ok i think i have insomnia . 0.775 -- but i might be suffering from temporary insomnia . 0.771 -- unfortunate to admit , but it s true ... well ... insomnia or go to bed ... i choose insomnia ! 0.745 -- insomnia , insomnia must sleep on it to think it over ... 0.742 -- i also think that s why i have insomnia . 0.738 -- i don t have insomnia . 0.733 -- im tired ... i think i have sleep insomnia or something . 0.705 -- i m probably overreacting , but i can t control my insomnia . 0.703 -- i think i may have narcolepsy ... oh well .

scala> query("horrible abortion", 10) adding horrible to vector adding abortion to vector

0.863 -- abortion ? 0.827 -- yes abortion ? 0.743 -- had an abortion . 0.730 -- they like abortions . 0.698 -- abortion : for or against it ? 0.657 -- i dont really believe in abortion . 0.628 -- a horrible horrible death . 0.620 -- no what do you think about abortion ? 0.619 -- abortion is wrong : in some cases , in others , no . 0.617 -- horrible ?

scala> query("gross fuckin miscarriage", 10) adding gross to vector adding fuckin to vector adding miscarriage to vector

0.705 -- it was fucking gross . 0.658 -- : nope ... gross 0.649 -- and hayleys like oh gross ! 0.639 -- fuckin . 0.638 -- so gross . 0.632 -- gross , i know . 0.631 -- gross , i d fucking kill him . 0.629 -- fuckin bitch . 0.628 -- shes fucking gross now with her blonde hair and her gross skinny ness . 0.626 -- gross .

scala> query("Nobody will hire me", 10) adding nobody to vector adding will to vector adding hire to vector adding me to vector

0.843 -- the other bad thing is people will still hire me , not too many will chance hiring my son 0.838 -- walgreens might hire me . 0.826 -- i really hope they hire me . 0.823 -- if anyone wants to go with me let me know . 0.819 -- if anyone wants to come , let me know ! 0.818 -- will someone help me ? 0.818 -- i also wanted to ask if they had any openings to hire me . 0.818 -- if anybody wants to do something give me a call ! 0.813 -- but i will always have someone who wants me . 0.812 -- maybe god will present him to me again someday and tell me that he is right for me .

scala> query("READY FOR SCHOOL", 10) adding ready to vector adding for to vector adding school to vector

0.889 -- getting ready for school and all . 0.808 -- horray for school . 0.805 -- im not ready for school . 0.804 -- i got ready for school . 0.791 -- i should get ready for school now . 0.770 -- oh well ... time to get ready for school tommorow ! 0.755 -- i have to get ready for school now , though . 0.747 -- today after school all my friends are getting ready for homecoming . 0.734 -- school for another week . 0.729 -- alright , well time for school .

scala> query("I like my new neighbors", 10) adding i to vector adding like to vector adding my to vector adding new to vector adding neighbors to vector

0.891 -- i like my friends . 0.873 -- i just like my jmu . 0.867 -- i really like my new things . 0.867 -- i like my friends and kelsey ahha . 0.859 -- i do it with my friends . 0.858 -- i feel like i have failed my friends for the month that i wasnt me . 0.858 -- i like seeing my friends though . 0.858 -- i like my sister 0.857 -- i love my neighbors , seriously . 0.857 -- i love it ... i like my swiffer wetjet too yes , i have my own , shut up .

scala> query("cheated", 10) adding cheated to vector

0.782 -- been cheated on ? 0.747 -- cheated on someone ? 0.745 -- cheated in a relationship . 0.726 -- : idk ever cheated ? 0.722 -- : yes been cheated on ? 0.698 -- : yep been cheated on ? 0.693 -- yup been cheated on ? 0.689 -- ever been cheated on ? 0.682 -- cheated on a test guilty . 0.681 -- yeah been cheated on ?

— Reply to this email directly or view it on GitHub https://github.com/berkeley-dsc/destress/issues/33#issuecomment-93933017 .

[davclark]

@geneyoo I don't have a problem with using mercury as a solr server. If you need https, I have a certificate set up for that - the best approach is to do forwarding via nginx.

That said, I've never done it. If you want to do this, let's be in the same place so I can be responsive as necessary.

If we continue this Solr discussion, we should probably move it to a different issue.