I've tried to replicate the exercise in the OP, but instead of focusing on topic title (which is what I did before), I went for the
whole merited post content. I has hopes that it would bring something meaningful, but I'm not convinced now, and actually prefer the results of the first approach (trending topics) to the later (merited posts), although the context is different.
The WordCloud I've created is for Meta. What I've done is count the number of occurrences all the words of all the Meta merited posts, then group manually words with a common structure (i.e. Merit and Merits) from the top of the list, and finally proceeded to perform a manual selection at my discretion of what I considered were relevant words (i.e. discard words like 'the', 'to', 'this' and so on).
The result for Meta is as follows:
It varies from the OP WordCloud for Meta, but I don’t see a clear relation to being merited by using a certain set of keywords. This goes to show, as I see it, that it’s not just the words that are necessary, but a proper construction and context. Of course these are not retrievable from just a word count, nor is the post’s overall tone and style.
The number of occurrences of the above words is as follows in the aggregate Meta awarded posts:
tag nTimes
---------------------------------------------
rank 5900
Post 9978
Merit 7631
thread 1924
give 4565
forum 2071
people 2062
system 2017
sMerit 1680
account 1304
spam 1764
members 1764
User 1999
think 1331
topic 1454
time 1082
board 1214
quality 891
need 811
know 801
receive 898
Bitcoin 665
trust 659
newbies 877
source 598
campaign 831
activity 533
read 465
help 454
signature 432
find 392
take 392
theymos 390
work 381
better 370
list 369
Jr 359
earn 358
Discussion 349
section 349
Hero 326
problem 319
idea 312
Legendary 299
bitcointalk 296
English 289
trying 286
rules 280
community 268
Please 268
person 267
example 265
understand 259
believe 255
bad 254
bounty 254
reason 254
negative 251
great 243
data 242
link 239
hope 238
unique 237
agree 234
different 234
OP 230
money 229
profile 228
local 226
change 224
create 218
opinion 214
check 212
question 201
meta 200
learn 197
I'm not sure it's a line of work worth pursuing...
