Jump to content
Sign in to follow this  
PhoenixSequence

Mind Reader - movement prediction software

Recommended Posts

I'm a software developer and as such created - Mind Reader.

Mind Reader's purpose is to query data in real-time to make a determination as to what will or will not most likely happen with a relatively high degree of probability.  

Mind Reader is designed to find and know anything of importance before it hits any main news outlets which might alter current markets. 

It works by interfacing directly with Twitter's databases and soon with other platforms such as Facebook. Which enables its users to know precisely what the public is thinking, again in real time, about anything of significance that's currently being of thought of, discussed or starting to trend - globally.

You start by entering any query into the input field and hit submit. XRP is the default for obvious reasons but you can remove that and enter any topic you like. 

You can query multiple keywords at once, and combine keywords in different ways. For example, to search for sentients with the words "cat" and "dog", enter: cat dog. To search for sentients with the words "cat" or "dog", enter: cat OR dog. To search for sentients with only the phrase "cat dog", enter: "cat dog". To search for sentients with the word "cat" but not "dog", enter: cat -dog.

Try entering "Bearableguy123", that will be fun. 

 

Click to begin - https://mindreader14.herokuapp.com/

I designed Mind Reader to be intuitive and easy to use.

At the top is a graph displaying what the world feels about the particular queried topic as well a returned sentiment score for each returned output. The sentiments themselves output below. 

I've also added a feature to export the returned sentiments and scores to an excel file to use the data in some third party software and reporting. 

I'm thinking if we all work together using this powerful tool and share what we find with one another (at least with me I hope lol) we can "know" what the world is thinking and ergo "predict" what's most likely going to occur. 

In this case, we can use it on XRP and know what's about to happen beforehand with it. 

The next phase will be to improve its user interface, finish adding connectivity to Facebook's databases and a few other sentiment data sources. Add an A.I. (machine learning) component that is specific to users log in. Giving it the ability to continuously query a selected topic and automatically find patterns within the sentiments and sentiment scoring. Thus making better predictions and plot the output data for those users. I also plan on making the URLs within the returned data clickable. Currently, you have to manually select them to open them.

Depending on how it does here, I'm considering customizing and licensing Mind Reader out to companies and organizations who would want to know what the world is thinking of them (the word on street) and any sentiments regarding their business or products they are selling.

As such, I thought this forum would be a great way to beta test it as well as provide an opportunity for me to give back. 

With Mind Reader the future is no longer unknowable.

Thank you.

Edited by PhoenixSequence

Share this post


Link to post
Share on other sites
18 hours ago, PhoenixSequence said:

It works by interfacing directly with Twitter's databases and soon with other platforms such as Facebook.

Reading the minds of the . . . will get nothing worth knowing . . . it´s just a thermometer in the a$$ of the social media bla-bla-blaaaaaaaaaa posters.
Sorry. But that is my rather Harsh Humble Opinion. I might be wrong.

Share this post


Link to post
Share on other sites
1 hour ago, zenkert said:

Reading the minds of the . . . will get nothing worth knowing . . . it´s just a thermometer in the a$$ of the social media bla-bla-blaaaaaaaaaa posters.
Sorry. But that is my rather Harsh Humble Opinion. I might be wrong.

Thanks for responding and I respect your opinion. However the responses and the results suggest otherwise. The best of luck to you.

Edited by PhoenixSequence

Share this post


Link to post
Share on other sites

I love seeing people DIYing these types of ideas - well done (I'm a developer myself).

The following is just random thoughts that are highly specific to XRP - I know your tool must be keyword matching as the input starts with a keyword...but I'm thinking I might even try and tackle the below ideas in terms of developing a sentiment analysis tool for XRP:-

  • build up profiles of Twitter users as you collect data of recent tweets - over time, you can see if an individual user is pro-XRP or anti-XRP.  You can then apply that knowledge to the sentiment analysis of recent tweets.  If a pro-XRP user posts a pro-XRP tweet, it's hardly surprising and it can be flagged as such.  A burst of new users tweeting about XRP would be very interesting - that would show XRP has maybe been mentioned on some mainstream media somewhere.
  • build up a profile for each hashtag and see how often they're used in positive and negative posts.  I'm guessing a hashtag like #XRPTheStandard is way way more likely to be used positively than in an ironic, negative way.  Having said that, it would be interesting to flag when anti-XRP users use pro-XRP hashtags - you could even flag a tweet as "possible ironic use of hashtag used" :)
  • look at mentions in tweets - are certain mentions used in positive tweets more than others? For example, in my anecdotal experience, pro-XRP tweets tend to mention @joelkatz way more than negative tweets would.  You can then profile mentions (over time) as something that might signify positive or negative sentiment
  • URL analysis.  Over time, you can build up a profile against each domain name in URLs posted on Twitter.  Is a particular domain name cited in pro-XRP tweets more than anti-XRP tweets?
  • for all of the above - manually teach the tool.  A human mind will detect irony and sarcasm so this manual teaching can really help make the tool more accurate.  For example, is "moon" used more for positive tweets than negative tweets? The only way to teach the system this is to on occasion go through recent tweets and manually flag tweets as positive or negative to teach the tool.  In fact, I think this occasional manual process can teach the tool over time all kinds of things (keywords, hashtags, URLs).  It doesn't need to be time consuming - I imagine having tickboxes next to each tweet (positive, negative, neutral) and you tick them (as admin, behind a login) to teach the system.

Just thinking out loud here.  I know some of the points above aren't trivial to implement.  I might have a go myself at developing such a tool as something to keep me occupied...

Share this post


Link to post
Share on other sites
On 1/19/2019 at 4:58 AM, PhoenixSequence said:

I'm a software developer and as such created - Mind Reader.

Mind Reader's purpose is to query data in real-time to make a determination as to what will or will not most likely happen with a relatively high degree of probability.  

Mind Reader is designed to find and know anything of importance before it hits any main news outlets which might alter current markets. 

It works by interfacing directly with Twitter's databases and soon with other platforms such as Facebook. Which enables its users to know precisely what the public is thinking, again in real time, about anything of significance that's currently being of thought of, discussed or starting to trend - globally.

You start by entering any query into the input field and hit submit. XRP is the default for obvious reasons but you can remove that and enter any topic you like. 

You can query multiple keywords at once, and combine keywords in different ways. For example, to search for sentients with the words "cat" and "dog", enter: cat dog. To search for sentients with the words "cat" or "dog", enter: cat OR dog. To search for sentients with only the phrase "cat dog", enter: "cat dog". To search for sentients with the word "cat" but not "dog", enter: cat -dog.

Try entering "Bearableguy123", that will be fun. 

 

Click to begin - https://mindreader14.herokuapp.com/

I designed Mind Reader to be intuitive and easy to use.

At the top is a graph displaying what the world feels about the particular queried topic as well a returned sentiment score for each returned output. The sentiments themselves output below. 

I've also added a feature to export the returned sentiments and scores to an excel file to use the data in some third party software and reporting. 

I'm thinking if we all work together using this powerful tool and share what we find with one another (at least with me I hope lol) we can "know" what the world is thinking and ergo "predict" what's most likely going to occur. 

In this case, we can use it on XRP and know what's about to happen beforehand with it. 

The next phase will be to improve its user interface, finish adding connectivity to Facebook's databases and a few other sentiment data sources. Add an A.I. (machine learning) component that is specific to users log in. Giving it the ability to continuously query a selected topic and automatically find patterns within the sentiments and sentiment scoring. Thus making better predictions and plot the output data for those users. I also plan on making the URLs within the returned data clickable. Currently, you have to manually select them to open them.

Depending on how it does here, I'm considering customizing and licensing Mind Reader out to companies and organizations who would want to know what the world is thinking of them (the word on street) and any sentiments regarding their business or products they are selling.

As such, I thought this forum would be a great way to beta test it as well as provide an opportunity for me to give back. 

With Mind Reader the future is no longer unknowable.

Thank you.

I really like this tool, its really cool.

Can only imagaine what it will be like in a year or two.

Well done, should be proud of your work there :) !

Share this post


Link to post
Share on other sites

Have you checked out IBM's Node Red project? It allows you to very easily create things like this in which you are streaming data from sources, and running it through various processing pipelines and then visualising it somehow. You have full access to IBM's Watson AI services, one of which is a sentiment analyser.

An example of building something with it here:

 

-Matt

Share this post


Link to post
Share on other sites
16 hours ago, mariusthegreat said:

I really like this tool, its really cool.

Can only imagaine what it will be like in a year or two.

Well done, should be proud of your work there :) !

Thank you. I really appreciate that.

Share this post


Link to post
Share on other sites
13 hours ago, hammertoe said:

Have you checked out IBM's Node Red project? It allows you to very easily create things like this in which you are streaming data from sources, and running it through various processing pipelines and then visualising it somehow. You have full access to IBM's Watson AI services, one of which is a sentiment analyser.

An example of building something with it here:

 

-Matt

Thank you. I'll take a look.

Share this post


Link to post
Share on other sites
16 hours ago, mariusthegreat said:

I really like this tool, its really cool.

Can only imagaine what it will be like in a year or two.

Well done, should be proud of your work there :) !

Thank you for the vote of confidence. I really appreciate that.

Share this post


Link to post
Share on other sites
On 1/19/2019 at 9:45 PM, 2ndtimearound said:

I love seeing people DIYing these types of ideas - well done (I'm a developer myself).

The following is just random thoughts that are highly specific to XRP - I know your tool must be keyword matching as the input starts with a keyword...but I'm thinking I might even try and tackle the below ideas in terms of developing a sentiment analysis tool for XRP:-

  • build up profiles of Twitter users as you collect data of recent tweets - over time, you can see if an individual user is pro-XRP or anti-XRP.  You can then apply that knowledge to the sentiment analysis of recent tweets.  If a pro-XRP user posts a pro-XRP tweet, it's hardly surprising and it can be flagged as such.  A burst of new users tweeting about XRP would be very interesting - that would show XRP has maybe been mentioned on some mainstream media somewhere.
  • build up a profile for each hashtag and see how often they're used in positive and negative posts.  I'm guessing a hashtag like #XRPTheStandard is way way more likely to be used positively than in an ironic, negative way.  Having said that, it would be interesting to flag when anti-XRP users use pro-XRP hashtags - you could even flag a tweet as "possible ironic use of hashtag used" :)
  • look at mentions in tweets - are certain mentions used in positive tweets more than others? For example, in my anecdotal experience, pro-XRP tweets tend to mention @joelkatz way more than negative tweets would.  You can then profile mentions (over time) as something that might signify positive or negative sentiment
  • URL analysis.  Over time, you can build up a profile against each domain name in URLs posted on Twitter.  Is a particular domain name cited in pro-XRP tweets more than anti-XRP tweets?
  • for all of the above - manually teach the tool.  A human mind will detect irony and sarcasm so this manual teaching can really help make the tool more accurate.  For example, is "moon" used more for positive tweets than negative tweets? The only way to teach the system this is to on occasion go through recent tweets and manually flag tweets as positive or negative to teach the tool.  In fact, I think this occasional manual process can teach the tool over time all kinds of things (keywords, hashtags, URLs).  It doesn't need to be time consuming - I imagine having tickboxes next to each tweet (positive, negative, neutral) and you tick them (as admin, behind a login) to teach the system.

Just thinking out loud here.  I know some of the points above aren't trivial to implement.  I might have a go myself at developing such a tool as something to keep me occupied...

Those are some very cool ideas. Thank you for the awesome feedback.

Share this post


Link to post
Share on other sites
9 minutes ago, DarkHorseXRP said:

I tried "Trump" ... and it's negative ...

Please....please....don't turn this political.  I'm practically begging.

Share this post


Link to post
Share on other sites
Sign in to follow this  

×
×
  • Create New...