Dear researchers,
I am looking for spatial-keywords Twitter or other datasets. I am familiar with Twitter APIs, but it does not give me what I want. I am searching for a website which has stored spatial-keywords data that contains geolocation information and keyword information . I am wondering how and where I will be able to find such data.
I would be grateful to hear your kind responses.
example：
"GeoLocation":"a:25.07926335;o:121.23692036" “keywords”：“Michigan Wisconsin muchos depressingo ”
Michael John Patrick
"GeoLocation":"a:25.07926335;o:121.23692036" “keywords”：“Michigan Wisconsin muchos depressingo”
What is the origin of those keywords? Twitter post metadata, or extracted from the message body by Twitter (or you)?
Michael John Patrick
extracted from the message body by Twitter,Remove useless words from Twitter, such as stop words, etc
I'd like to be able to count the word frequency to show the popular words, but there will be significantly more stop words
Thanks for your answer
What is your definition of the term 'keyword'. For instance, the term usually (for me ) references to a taxonomy ( https://en.wikipedia.org/wiki/Taxonomy ) or a typology ( https://www.britannica.com/science/typology ) for a given domain - for instance for Geomorphology, they are established and recognized by a specialized community of users.
Twitter content, being a public unstructured random set of text, wouldn't have au priori established keywords (according to my definition ), although it does have 'hashtags' which are generated by a user, and through use can become a sort of informal ad hoc taxonomy called a 'folksonomy' for participants in a thread or following.
The Twitter post meta data contains geographical information: https://developer.twitter.com/en/docs/twitter-api/v1/data-dictionary/object-model/geo this is obtained automatically from the cell phone GPS, IP addresses, the user profile, etc. , but my understanding is that you are referring to the noun phrases taken from the body text of the content ( https://business.twitter.com/en/help/campaign-setup/campaign-targeting/keyword-targeting.html ) - the marketing and advertising meaning of 'Keywords' - these are really just tokens for matching like the hashtags. (https://www.tokenex.com/blog/ab-what-is-nlp-natural-language-processing-tokenization/ ). Unlike the formal definition of Keyword which have meaning in an established structure, the Twitter keywords do not - 'Cheyenne' could be a female name, a city in Wyoming, a perfume, etc.
Geographic location names, 'Toponyms' are tricky - for well established entities, there are utilities like the National Geospatial Intelligence Agency's Geographical Name Server (GNS) at https://geonames.nga.mil/geonames/GNSHome/welcome.html (for International toponyms ) and the USGS Geographic Names Information System (GNIS) at https://www.usgs.gov/us-board-on-geographic-names/domestic-names - but disambiguation of a Twitter 'keyword' as a place name would fairly difficult without context - see graphic - 'maria' returns 8273 possibilities.
The IBM Watson Natural Language Understanding Text Analysis ( https://www.ibm.com/demos/live/natural-language-understanding/self-service/home ), if there is enough surrounding text to derive some sort context, will return a ranked probability list.
Without a clear understanding of your goal, that's the best I can do for now. GeoNLP is a vast academic topic.