The Basics
Xmo.ai is a free tool to analyze emotions, sentiment, and other linguistic stats in X content. You may search by username, thread URL, or any valid X search query below. In the first case, all matching posts (up to the 20 most recent posts) on the user's timeline will be analyzed (including quotes/reposts); in the second, the OP will be analyzed separately from replies (1 level deep; stats averaged over up to the top 10 replies - "top 10" here means the 10 replies that are shown first when navigating to this thread). In the final case, averaged stats are shown for the most recent matching tweets (up to 20), regardless of who the author is. Please note that in general, reposts/quotes are not counted as they are not part of the author's original content. The exception to this is when a users wants to analzye by thread URL - in this case, if the OP contains a repost/quote, that content will read as if it was written by the user quoting/reposting it, but in other cases (replies), quotes/reposts are not included. I made this decision primarily because in very popular threads, the replies typically include a ton of spam that usually includes quotes/reposts. Eventually, all of this (the # of posts analyzed, how to treat quotes/reposts, etc) will all be customizable by the user performing the search.
In general, the purpose of Xmo.ai is to allow people to see how people react to content on X, based on what can be learned from the language of their posts. Specifically, analyzing a thread in this way allows one to see sentiment/emotions of the OP as well and the replies separately, in order to see how people are reacting to the post. Note that this works for anything with a valid X url - sometimes someone replies in a particular way that attracts attention and nearly goes viral itself. You can also analyze such content by simply pasting the appropriate X URL.
Tech Info
Xmo is built on a standard LAMP (Linux/Apache/MySQL/PHP) stack and uses a privately self-hosted instance of the fantastic open-source project Nitter, an alternative front-end for X (fka Twitter). The Twinword API is used for most tasks related to natural language processing and a few other open-source tools and APIs are used as well (Quickchart, PHPlot, FreeDetector.AI, Unavatar). If you're curious about the particulars, feel free to get in contact with me. If you're a tech person, you're likely wondering, "Wouldn't this be a whole lot easier using the official X developer API?" The answer is a resounding yes. Unfortunately, the official API "tier" that would have been useful for this project was jacked up from $0/mo (yes, it was free) to $100/mo following the Twitter acquisition. This is obviously unreasonable for non-commercial projects that are a hobby (side note: I was forced to shut down 2 entirely separate projects of mine because of this). Feel free to complain to @XDevelopers about this, I certainly have... and let's just hope that alternative front-ends like Nitter continue to function for the foreseeable future. No logs of queries are kept, and you do not need to be logged into your X account or sign up for another account at Xmo.ai to use our service - so feel free to search knowing your privacy is protected.
Future Plans
Yes, I do have some plans to expand this beyond just X content, but that's all I'm going to say about it publicly for now. You can get in contact with me if you'd like to discuss it privately.
Contact Info
If you'd like to get in touch me, you can send an email to hi[at]xmo.ai or find me on X (@MagicMirrorAI).
Xmo.ai is a free tool to analyze emotions, sentiment, and other linguistic stats in X content. You may search by username, thread URL, or any valid X search query below. In the first case, all matching posts (up to the 20 most recent posts) on the user's timeline will be analyzed (including quotes/reposts); in the second, the OP will be analyzed separately from replies (1 level deep; stats averaged over up to the top 10 replies - "top 10" here means the 10 replies that are shown first when navigating to this thread). In the final case, averaged stats are shown for the most recent matching tweets (up to 20), regardless of who the author is. Please note that in general, reposts/quotes are not counted as they are not part of the author's original content. The exception to this is when a users wants to analzye by thread URL - in this case, if the OP contains a repost/quote, that content will read as if it was written by the user quoting/reposting it, but in other cases (replies), quotes/reposts are not included. I made this decision primarily because in very popular threads, the replies typically include a ton of spam that usually includes quotes/reposts. Eventually, all of this (the # of posts analyzed, how to treat quotes/reposts, etc) will all be customizable by the user performing the search.
In general, the purpose of Xmo.ai is to allow people to see how people react to content on X, based on what can be learned from the language of their posts. Specifically, analyzing a thread in this way allows one to see sentiment/emotions of the OP as well and the replies separately, in order to see how people are reacting to the post. Note that this works for anything with a valid X url - sometimes someone replies in a particular way that attracts attention and nearly goes viral itself. You can also analyze such content by simply pasting the appropriate X URL.
Tech Info
Xmo is built on a standard LAMP (Linux/Apache/MySQL/PHP) stack and uses a privately self-hosted instance of the fantastic open-source project Nitter, an alternative front-end for X (fka Twitter). The Twinword API is used for most tasks related to natural language processing and a few other open-source tools and APIs are used as well (Quickchart, PHPlot, FreeDetector.AI, Unavatar). If you're curious about the particulars, feel free to get in contact with me. If you're a tech person, you're likely wondering, "Wouldn't this be a whole lot easier using the official X developer API?" The answer is a resounding yes. Unfortunately, the official API "tier" that would have been useful for this project was jacked up from $0/mo (yes, it was free) to $100/mo following the Twitter acquisition. This is obviously unreasonable for non-commercial projects that are a hobby (side note: I was forced to shut down 2 entirely separate projects of mine because of this). Feel free to complain to @XDevelopers about this, I certainly have... and let's just hope that alternative front-ends like Nitter continue to function for the foreseeable future. No logs of queries are kept, and you do not need to be logged into your X account or sign up for another account at Xmo.ai to use our service - so feel free to search knowing your privacy is protected.
Future Plans
Yes, I do have some plans to expand this beyond just X content, but that's all I'm going to say about it publicly for now. You can get in contact with me if you'd like to discuss it privately.
Contact Info
If you'd like to get in touch me, you can send an email to hi[at]xmo.ai or find me on X (@MagicMirrorAI).