User testimonials

[kbenoit]

Please use this issue to add your testimonials, experiences, feedback, etc. We would love to hear from you.

Updated consent notice 2017-10-06:

Note that in leaving a testimonial, you provide consent for us to reproduce it on our website http://quanteda.io or in other promotional materials for the package, should we choose to do so. We will attribute this only to your GitHub account, which means in effect that it will not add any additional information than what you will have already posted to this issue (which means the comment and your GitHub username).

Thanks very much!

-- @kbenoit and the whole quanteda team

[BobMuenchen]

The quanteda package has become the starting point for all my text analysis projects in R. This is due to its superb ability to prepare the data for analysis regardless of the method one might prefer. Single words are notoriously ambiguous, and quanteda can help find phrases (ngrams), and convert them to single terms before removing any stop words that they may contain. When the data is prepared, quanteda's convert function makes it easy to transfer the data to a wide range of other packages such as topicmodels or lsa.

quanteda also contains functions to create dictionaries that automate the identification of topics, and it can import and apply the wide range of existing dictionaries that exist in several formats.

[cbpuschmann]

Quanteda is an excellent resource for both research and teaching than complements R in a way that is invaluable to me -- only switching to Python would offer comparable benefits. It is far superior to related packages (e.g. tm) and so well documented that I use it centrally when teaching text mining.

I hope that the package's rapid development will continue, with the integration of further exciting features, such as integration with the package topicmodels, supervised ML, easy application of dictionaries/thesauri etc in the future. Quanteda constitutes a major step forward for social scientists interested in text mining.

[jofrerocabert]

Quanteda has been a greatly valuable tool for my research. Text analysis is a rapidly expanding field with new applications appearing rapidly in the social sciences. This package simplifies many of the tasks and allows me to implement more sophisticated classifiers for a more detailed analysis.

[kbenoit]

Thanks so much @BobMuenchen @cbpuschmann and @jofrerocabert !

[cschwem2er]

Quanteda is my favorite toolbox for natural language processing and its so nice to see how much progress you made within the last year. As of now, the package can even be used as a complete NLP backbone for interactive web apps.
In the future I'd love to see better workflow integration with the tidyverse packages and supervised models.
Thank you so much Ken and the whole quanteda team for creating yet another awesome package! =)

[DenisaKost]

I am a political scientist, without a background either in computer science or natural language processing. I have used quanteda to analyse a large volume of textual data in four different languages from the Balkans. This user-friendly package has allowed me to apply quantitative text analysis to make a major theoretical breakthrough in my field of study, which is post-conflict justice and peacebuilding. I have particularly benefited from quanteda’s meticulous documentation, which I consulted regularly to carry out my analysis.

[lmkirvan]

This package is amazing! I use it on a daily basis to perform a large number of text analysis tasks. The responsiveness of Ken and the rest of the team is simply amazing. They really pay attention to and help their users. Others at my federal agency now use it frequently for text analysis in R too (once others saw how clean my code was and how fast it ran it didn't take much convincing).

I personally use it to understand complaints submitted by hundreds of thousands of consumers. The whole agency benefits from being able to understand more text from our stakeholders faster. Kudos to the whole team.

[kbenoit]

Thanks so much @lmkirvan for the kind words, and keep those feature requests and bug reports coming - we welcome the feedback as a way to improve the package!

[andremog]

I participated in one of the text analysis workshops about the quanteda package in April 2017. The package offers a wide range of functions applicable to many use cases. Even though I am relatively new to text analysis methodologies in general, the package has a very intuitive structure for its commands, making them easy to understand. Its powerful commands allow complex analyses with just a few lines of codes, building and replicating results quickly (as all commands run quite fast). The development team is open to user requests (e.g. requests for new features) and willing to help address difficult use cases. Thanks a lot to the team!

[kbenoit]

Thanks @andremog and glad you found the workshop useful!

[SeraphineM]

Quanteda offers a range of powerfull tools and is currently the best package available for doing quantitative text analysis in R. Apart from this, what I particularly like about the package is the easy implementation of dictionaries in various formats. Noteworthy is here also the impressively quick and efficient support of the developers in case of bugs: keep up the good work and thank you!

[kbenoit]

Thanks @SeraphineM !

[lgreski]

Hello Ken. I recommend quanteda to students in the Johns Hopkins University Data Science Specialization Capstone project, and include an analysis of the runtime for creating various sized ngrams for the Heliohost Corpus.

regards,

Len

[kbenoit]

Fantastic! Thanks for the endorsement and vote of confidence @lgreski.

[Pedromoisescamacho]

quanteda is the most understandable and faster text mining package that I have tried. In comparison with TM (the most popular package), quanteda is lighting fast and the workflow is more logical.

[johnzfeng]

Quanteda is a great tool to my work. Compared to TM (the most popular text analysis package in R), RTextTools and others, Quanteda is easier to use and runs faster. Kbenoit also responds to issues promptly. I am looking forward to the new Quanteda features he is building.

I recommend Quanteda to all levels of text miners.

[kbenoit]

@johnzfeng @Pedromoisescamacho @SeraphineM @andremog
Hope you’ve been enjoying our improvements to quanteda!

We’re revamping the http://quanteda.io website and may use a few testimonials. Would you mind if we used yours from this issue? (I've emailed the others directly but did not find your addresses.) We did not have a consent notice when you posted the comment, so I wanted to ask your permission first. See the updated notice above for details.

Feel free to reply here with a 👍 to this comment to give assent, or if you prefer, reply to me directly at kbenoit@lse.ac.uk.

Thanks!

[cbpuschmann]

👍

[Pedromoisescamacho]

It would be an honor

…

On Oct 6, 2017 05:27, "Cornelius Puschmann" ***@***.***> wrote: 👍 — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#461 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AKFPeLIyPgF0BAoEB-7vrY-TmAXSSTerks5spfJngaJpZM4LjmZv> .

[andremog]

@kbenoit No problem at all. Thanks!

[johnzfeng]

Thanks for the update, kbenoit. Quanteda is a very convenient and easy to use text analytic tool. I am glad that you are making more progress and I am looking forward to using the updated function.

[jansenjoost]

As a PhD candidate in Sociology I do text analysis on a daily basis and the quanteda package is, by far, the most convenient, consistent to work with, and comprehensive package I've worked with. @kbenoit and his co-developers are constantly busy trying to extend quanteda's functionality. Moreover, they provide excellent and really quick support to all of your questions (e.g. on the Stack Overflow platform), as simple or difficult as they might be. I'll defenitely spread the word amongst my colleagues.

[dkholo]

I am on the other hand has already initiated the on-boarding process of quanteda to Power BI cloud service. So soon it will be possible to publish beautiful quanteda dashboards and share them globally. My encounter with quanteda started as a part of my ML course project, I was trying to detect fluctuations of negativity in the large sets of documents over time. Initially I was planning to use tm package, but this package could not handle the work load. I was completely blocked and was considering alternative options for my project. Then my professor suggested to use quanteda. I was quickly unlocked and was able process 50 million rows of text on a small laptop. I was easily able to complete my project. I can see a great potential in this project. With Power BI integration the data scientists of the world will have a chance of utilizing the quanteda capabilities on dynamic text data sources and utilizing the text analytics beyond the limits of lexical studies. I personally depend on xray chart evolution and capability expansion. I plan on building real time text to data dashboards in the near future.

[MatthewLacombe]

I'm a PhD candidate in political science and Quanteda has been a wonderful resource for me. It's very user-friendly, allows for quick and flexible preparation of data, includes a comprehensive set of functions in a single package, and runs fast. Further, Ken and the rest of the team are incredibly responsive to inquiries and are very helpful in their replies. I'll happily recommend to my colleagues!

[enrico200165]

I am senior in IT but junior in NLP, came to quanteda after noticing that nearly everybody involved in an NLP exam was abandoning other packaging due to meeting functional or performance limitations, and adopting quanteda, about which feedbacks were only positive.
Personally I was impressed by timing quality (and also patience) of replies by package authors to some newbie questions I asked.
I am not yet very familiar with the design but it seems to be based on few clear principles applied as cleanly as possible.

[guilhermeparreira]

I started using quanteda due to a limitation of tm in dealing with documents in Portuguese (specially in stemming). Moreover I asked for an extra argument for textplot_networkfunction, and they just did!
So cool!! I am really enjoying this package!

[gfzb]

We are a young company based in Zurich and specialize in policy evaluation. Wherever possible, we try to leverage the potential of data science methods in our projects. quanteda is our first choice when dealing with large amounts of text, because it is fast and stable, it offers many possibilities to quickly find patterns and it provides excellent interfaces for further analysis. Our customers are always impressed how much insight can be gained from large amounts of unstructured text data in short time. Without quanteda our job would be less easy and definitely less fun.

[kbenoit]

Thanks so much @gfzb we love to hear that sort of feedback!!

[AmyLei96]

I have been doing a comparative analysis of R text mining packages and quanteda by far has had everything I've needed to test. For someone who has no prior experience with text analysis, this package has made it super fun and easy for me to explore this field.

[snsoroka]

Just a quick note to say how much I'm enjoying using quanteda. It's become my first stop for content analysis, for both dictionary- and supervised-learning-focused projects. It's made data management a lot simpler, because I don't have to move back and forth between different packages with different data-formatting rules. Because it draws together so many different techniques, it facilitates a more comprehensive analysis of text as well -- it's now very straightforward to compare results from several different approaches. Both my research and teaching on content analysis are increasingly based on quanteda. Thanks!

[abresler]

Quanteda is THE swiss army knife for NLP in R.

It combines everything one needs to perform state of the art text analysis.

• Speed

• Clean API

• Out-of-the box visualizations

• Other NLP API wrappers

Quanteda has played a central role in a major effort to parse large text blocks in order to create word tags, find keywords and text collocations. The pipeline has helped solve a major bottleneck, how to create matches without structured data.

This process is being to assist both government and commercial customers to put their text data to work.

This package truly does it all, its fast, comprehensible and extendable and I look forward to scaling out additional use cases that leverage this amazing package to help make more out of text data.

[promothesh]

I am prepping to teach Text Analytics for the first time and Quanteda is incredibly useful. What a resource, thank you Ken Benoit and team for this amazing package.

[yadevinit]

In early 2020, I've shared following message with Ministry of Education of India (MHRD) and others. (This Ministry was earlier known as MHRD or Ministry of Human Resource Development.) The Project RampWalkAnyDifficulty (migrated from discontinued-website https://notebooks.azure.com/yadevinit/projects/rampwalkanydifficulty) referred to has used Quanteda and Structural Topic Modeling (STM). And around 2020-March, Ken had helped me deal with an issue over Github, and promptly. I wish to thank you (and related unsung heroes and heroines :-) for Quanteda and STM and sharing it the way you did.

"Ramp Walk" any (Occipital) Difficulty

Synopsis

MHRD has expressed a "sankalp" (commitment) of giving India her demographic dividend; that's reflected in the Vision and Objectives of institutions such as the NTA. While much of the world locked down facing the COVID-19 pandemic, Union HRD Minister launches AI-powered mobile app for mock tests of JEE Main, NEET 2020, other competitive exams. That was after MHRD webinar with students from over a million about to face competitive exams such as JEE-Main.

Concurrently and like never before, this Project extended Project Occipital that was shared with MHRD and others earlier, revised the discovery of (JEE) topics using AI/ML, and re-estimated the topic-wise difficulty. For example: the Incidence Rate of right answers rightIR rises 6.5 times if a unit increase in prevalence is done of Maths topic "Differential Equations" instead of Physics topic "Electrostatics; Thermodynamics". (rightIR is the proportion of examinees who get a related question right.) Put simply, 6.5 times as many examinees find (questions from) this Physics topic difficult as that Maths topic.

Right now, NTA is well poised to leverage this Project: choose which topics to improve, and bridge related teaching-learning gaps. Would MHRD like to (a) invite NTA to do so and (b) lead the invitation to well wishers of India to offer projects for extraordinary teaching and learning, as called upon by Project PeTaL? By also (unlearning and) re-learning pre-requisite topics with the near-15-year-old students, what's possible is the next orbit of India's impact for the world, consistent with India's commitment for PISA 2021. This author invites you to consider and act now while you read ahead at Project RampWalkAnyDifficulty. There's something in it for (student) learners and others too.

(If you're new to such a URL link, you have to only scroll to read the document page there after clicking the link and launching it in a web browser. By mistake, some click at various places and end up downloading Project .zip or other files.)

Regards,
--- Vinit Kaushik

[kbenoit]

@juliajung11 you should make sure you are using the latest versions of everything, updating your R and the packages to their latest versions. I'm going to delete your comment from this thread once you verify that this worked, since this is an issue for testimonials, not problem reports.