🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Please use tokens_select() instead of kwic() to extract context words using #quanteda. Too many people are confused. blog.koheiw.net/?p=1854

Sorry that I didn't make this clear earlier!

安部政権とネタニエフ政権下での安全保障問題に関する新聞報道と首相支持率についての論文をオープンアクセスで公表しました。アジア言語（日本語とヘブライ語）の文書の高度なテキスト分析が簡単になった証拠としてみてください。#quanteda #LSS blog.koheiw.net/?p=1866

Encyclopedia entries on text analysis from fresh perspectives blog.koheiw.net/?p=1938 Thank you, Andrea Ceron

Please see the latest Geopolitical Threat Index (GTI) on an interactive web app: blog.koheiw.net/?p=1951 The score for #ukraine has skyrocketed in 2022 (as expected). The app is created using #quanteda, #LSX and #shiny in #R. The original research is with P. Trubowitz at LSE Phelan US Centre.

Belatedly, I published instructions on how to perform Latent Semantic Scaling (LSS) using the LSX package. I hope this helps: koheiw.github.io/LSX/articles/p…

Users should actively participate in discussions on open source development. Otherwise, the project stall and eventually die!

Very much looking forward to presenting part of my PhD research at the SODAS, Copenhagen (Bluesky: @cphsodas.bsky.social) Data Discussion next week! Thanks so much HjalmarCarlsen @[email protected] for the invite! Details: sodas.ku.dk/events/sodas-d…

I am developing the next version of #quanteda. One big change would be making tokens object XPtr-based to process larger data more efficiently. Please try the prototype and tell me how you like it: github.com/quanteda/quant…

We can estimate emotions of words and emojis accurately using #LSS and #quanteda. If you are interested, please read our article in the J of Medical Internet Research: jmir.org/2023/1/e44965/. This is my best shot in analysis of social media.

Having trouble identifying topics of sentences in large corpora? Use Distributed Sequential LDA implemented in the seededlda package. #textanalysis #nlp #rstats

日曜日の日本メディア学会の研究会で約束したとおり、種語の選び方について説明したページを作りました。koheiw.github.io/LSX/articles/p…

New #openaccess publication in Research & Politics. Kohei Watanabe and I present a new method to estimate temporal focus in text without Training data that only uses a low number of commonly used verbs and their inflections. journals.sagepub.com/doi/10.1177/20…

Our semantic temporality analysis recognises various temporal features beyond the tense of verbs such as adjectives, adverbs and lexical aspects. You don't need to use POS tagger or training data because it is based on semi-supervised algorithm!

LSX v1.4.0 has been released with a new visualization function. The plot highlights words in an LSS model about security threats with different colors depending on their association with China, North Korea, Iran or Russia blog.koheiw.net/?p=2105 #R #quanteda

Does anyone know how to link to C++ header files (installed using homebrew) in R package on MacOS? #Rstat #quanteda #macOS github.com/quanteda/quant…

You can make your R scripts for large-scale text analysis 2x faster and 3x more memory efficient using the external pointer tokens object (tokens_xptr) in new quanteda blog.koheiw.net/?p=2077 #RStats #quanteda

In topic modeling, it is important to adjust parameters for topic sizes. If you don't know how, read this post and try the new algorithm: blog.koheiw.net/?p=2191 Looking forward to your feedback. #LDA #rstats #quanteda

If you think the number of topics, k, is the only important parameter for topic models, you need to read this post and the research paper. I created a new model to optimize the Dirichlet priors to analyze imbalanced corpus more accurately. bsky.app/profile/koheiw…

A few days ago, I received an email from a researcher asking if text analysis is becoming irrelevant because of AI... please read my post "AI products and text analysis methods" blog.koheiw.net/?p=2254 #quanteda #rstats