Sentiment Analysis using the new Extract Sentiment operator

sgenzer

This has been a long time coming..."one operator to rule them all" for sentiment analysis. Introducing the Extract Sentiment operator from Operator Toolbox 2.0:



Parameters are very straightforward - choose one from a list of four possible methods (two commercial, two open source) and just execute. Extract Sentiment will give you a new labeled score attribute.





It's really that simple. There is anice tutorial processthat you can check out as well.

Enjoy!

Scott

kayman

Love it, Vader is a great yet underestimated library!

Telcontar120

@sgenzerthe link to the tutorial process in the OP appears to be broken.

sgenzer

@Telcontar120well how the heck did that happen?Ok hopefully this will work now. Thx.

Telcontar120

Yep,@sgenzerthat link works great now, thanks.
However, I think there is a problem with the operator---it doesn't seem to take the connections that are defined in the repository. At least it won't accept mine for Aylien or Meaningcloud.@mschmitz有什么特别的,这里会发生什么?The connections are visible when I click the "?" icon in the parameters but it won't seem to select them in the dropdown (which is blank and stubbornly remains so).

MartinLiebig

@Telcontar120

since both the Aylien and the MeaningCloud extensions aren't open source, i could not use their connection. You need to setup a new connection of type Aylien (Toolbox) or MeaningCloud (toolbox) respectivly.

Best,

Martin

Telcontar120

Ah, got it. I see that as a different connection option now. Perhaps you should add that clarification to the operator description? It says you need an API connection defined, but people who already have a connection of those types defined will probably think they don't need to add them again specifically for the Toolbox operator (like I did).

MartinLiebig

Hi@Telcontar120,

actually@sgenzerwrote the documentationSo all blame on him!

Serious note: of course. i will add it tomorrow, but we will most likely not do a release for it.

BR,

Martin

sgenzer

As my father-in-law says all the time:

No good deed goes unpunished.

Rojan

Hi Sgenzer, thank you for the post, I did exactly as above. Can you help me with interpretation of the results if there are any already, kindly help me find it? Else, once we run it we get the file with scores, that shall we use for interpretation and how to draw conclusions out of them. Than you in advance.

sgenzer

hi@RojanI'd strongly recommend looking at the tutorial processes with this operator (see Help panel)

jajderian

Hi! This looks like a very useful plugin, I'm keen to use it. Is there any more info about the actual types of models included? Eg are the underlying models decision trees, naive bayes, clustering algorithms? I'm planning on using this in an assignment and would like to add more detail. Plus it's always fun to know what's going on under the hood. Cheers

[Deleted User]

@mschmitz,@sgenzer

Very good documentation because it is very short and I can translate it easily

Thank you@sgenzer,短文章
mbs

HeikoeWin786

@sgenzer

Hello Sgenzer,

我可以知道为什么ther do we need to perform any text pre-processing before using this operator? .

thanks

sgenzer

hi@HeikoeWin786hmm you don't have to, but it may be a good idea depending on your use case.

SG