Yep, sounds like it. To give you an example of what I’ve done in the past:
* Azure Function set up on a timer every 15 minutes * Function scrapes any tweets related (via a bunch of logic) to our client’s organisation and/or any of its primary competitors * For each tweet, first I make a call for language detection. I only keep English language ones. (I found one there is a Brazilian supermarket with the same name as one of our clients so I need to exclude all the Portuguese discussion, etc.) * I also request keyword analysis, and sentiment analysis * I stick all of this in a DB * Analytics is built above that. Has proved very useful. For example, CEO can now see word clouds of what’s being discussed (not just Twitter but including it). Not long back, he was about to walk into a press conference where he was talking about their “green” investment strategies, and until he saw it in the dashboard, he was unaware he was about to be challenged about some of their specific investments. Very happy to have been forewarned. Similar deal with another client. We now are able to perform really interesting monitoring of call centre staff, by adding transcription into the mix. For example, in investment areas, they are not allowed to say things like “I recommend…”. Now we know if they do, and if so, what on earth did they say next? There are a whole bunch of things they aren’t allowed to say. There are also things they should say. We check for those as well. Which staff do the best job of removing the heat out of calls as they progress, etc.? The irony in this sort of thing is that if you’d asked the company board members (and even the IT management) about whether AI could apply to them, they’d have told you no. It’s all a matter of being able to imagine what’s possible. Regards, Greg Dr Greg Low 1300SQLSQL (1300 775 775) office | +61 419201410 mobile│ +61 3 8676 4913 fax SQL Down Under | Web: www.sqldownunder.com<https://nam06.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.sqldownunder.com%2F&data=02%7C01%7Csspahelp%40microsoft.com%7C1f0ea4d6b97e4d897f3708d666d1e890%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C636809449091516274&sdata=SLHeEGAMmWUY5YIwcC4oAPYr%2F9RIZdi4MNASsdzwX2I%3D&reserved=0> |http://greglow.me<https://nam06.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgreglow.me%2F&data=02%7C01%7Csspahelp%40microsoft.com%7C1f0ea4d6b97e4d897f3708d666d1e890%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C636809449091526278&sdata=IU8tnAITCjBxWafi3A9XpO9lF3PIwZJ8ad3t36lnxvs%3D&reserved=0> From: ozdotnet-boun...@ozdotnet.com <ozdotnet-boun...@ozdotnet.com> On Behalf Of Greg Keogh Sent: Tuesday, 9 February 2021 1:50 PM To: ozDotNet <ozdotnet@ozdotnet.com> Subject: Re: Text Analytics I’ve always included it in the REST based call. I’ve not seen any option for giving it a URI from which to retrieve the text. I’ve been feeding in tweets, and transcriptions of phone calls but haven’t run into size limits. We’ve been doing keyword analysis, sentiment, language detection, etc. I’m not sure that feeding in large volumes of text would be all that useful. For example, even when doing call transcriptions, I’m looking at sentiment at different parts of a call. Key phrases retrieved from gigabytes of text wouldn’t seem that useful to me. Hi GL, from what you say, maybe I misunderstand how it's supposed to be used?! Or is it the wrong tool? My colleague has megabytes of collected tweets on subjects like US politics and Covid and is looking for trends in sentiment and frequency of keywords like 'Trump" "impeach" "vaccination" etc. He is generating attractive charts of this sort of thing, but by a rather tedious process at the moment and I was hoping the Azure Text Analytics could make his task a breeze. I was going to run an experiment by feeding in the whole text of The War of The Worlds<https://www.gutenberg.org/files/36/36-0.txt> and looking for key phrases and major names. That's when I realised it didn't accept input the way I expected. or it doesn't work the way I expect. Hmmm... Greg K
