Help by editing my question title and tags is greatly appreciated!
Sometimes one participant in my corpus of "conversations" will refer to another participant using a nickname, usually an abbreviation or misspelling, but hereafter I'll just say "nicknames". Let's say I'm willing to manually tell my software whether or not I think various possible nicknames are in fact nicknames, but I want software to come up with a list of possible matches between the handle's that identify people, and the potential nicknames. How would I go about doing that?
Background on me and then my corpus: I have no experience doing natural language processing but I'm a competent data analyst with R. My data is produced by 70 teams, each forecasting the likelihood of 100 distinct events occurring some time in the future. The result that I have 70 x 100 = 7000 text files, containing the stream of forecasts participants make and the comments they include with their forecasts. I'll paste a very short snip of one of these text files below, this one had to do with whether the Malian government would enter talks with the MNLA:
02/12/2013 20:10: past_returns answered Yes: (50%)
I hadn't done a lot of research when I put in my previous
placeholder... I'm bumping up a lot due to DougL's forecast
02/12/2013 19:31: DougL answered Yes: (60%)
Weak President Traore wants talks if MNLA drops territorial claims.
Mali's military may not want talks. France wants talks. MNLA sugggests
it just needs autonomy. But in 7 weeks?
02/12/2013 10:59: past_returns answered No: (75%)
placeholder forecast...
http://www.irinnews.org/Report/97456/What-s-the-way-forward-for-Mali
My initial thoughts: Obviously I can start by providing the names I'm looking to match things up with... in the above example they would be past_returns and DougL (though there is no use of nicknames in the above). I wouldn't think it'd be that hard to get a computer to guess at minor misspellings (though I wouldn't personally know where to start). I can imagine that other tricks could be used, like assuming that a string is more likely to be a nickname if it is used much much more by one team, than by other teams. A nickname is more likely to refer to someone who spoke recently than someone who spoke long ago, or not at all on regarding this question. And they should be used in sentences in a manner similar to the way the full name/screenname is typically used in the corpus. But I'm interested to hear about simple approaches, as well as ones that try to consider more sophisticated techniques.
This could get about as complicated as you want to make it. From the semi-linguistic side of things, research topics would include Levenshtein Distance (for detecting minor misspellings of known names/nicknames) and Named Entity Recognition (for the task of detecting names/nicknames in the first place). Actually, NER's worth reading about, but existing systems might not help you much in your domain of forum handles and nicknames.
The first rough idea that comes to mind is that you could run a tokenized version of your corpus against an English dictionary (perhaps a dataset compiled from Wiktionary or something like WordNet) to find words that are candidates for names, then filter those through some heuristics (do they start with the same letters as known full names? Do they have a low Levenshtein distance from known names? Are they used more than once?).
You could also try some clustering or supervised ML algorithms against the non-word tokens. That might reveal some non-"word" tokens that often occur in the same threads as a given username; again, heuristics could help rule out some false positives.
Good luck; sounds like a fun problem - hope I mentioned at least one thing you hadn't already thought of.
Related
I've been working on a sentence transformation task that involves paraphrase identification as a critical step: if we are confident enough that the state of the program (a sentence repeatedly modified) has become a paraphrase of a target sentence, stop transforming. The overall goal is actually to study potential reasoning in predictive models that can generate language prior to a target sentence. The approach is just one specific way of reaching that goal. Nevertheless, I've become interested in the paraphrase identification task itself, as it's received some boost from language models recently.
The problem I run into is when I manipulate sentences from examples or datasets. For example, in this HuggingFace example, if I negate either sequence or change the subject to Bloomberg, I still get a majority "is paraphrase" prediction. I started going through many examples in the MSRPC training set and negating one sentence in a positive example or making one sentence in a negative example a paraphrase of the other, especially when doing so would be a few word edit. I found to my surprise that various language models, like bert-base-cased-finetuned-mrpc and textattack/roberta-base-MRPC, don't change their confidences much on these sorts of changes. It's surprising as these models claim an f1 score of 0.918+. The dataset is clearly missing a focus on negative examples and small perturbative examples.
My question is, are there datasets, techniques, or models that deal well when given small edits? I know that this is an extremely generic question, much more than is typically asked on StackOverflow, but my concern is in finding practical tools. If there is a theoretical technique, then it might not be suitable as I'm in the category of "available tools define your approach" rather than vice-versa. So I hope that the community would have a recommendation on this.
Short answer to the question: yes, they are overfitting. Most of the important NLP data sets are not actually well-crafted enough to test what they claim to test, and instead test the ability of the model to find subtle (and not-so-subtle) patterns in the data.
The best tool I know for creating data sets that help deal with this is Checklist. The corresponding paper, "Beyond Accuracy: Behavioral Testing of NLP models with CheckList" is very readable and goes into depth on this type of issue. They have a very relevant table... but need some terms:
We prompt users to evaluate each capability with
three different test types (when possible): Minimum Functionality tests, Invariance, and Directional Expectation tests... A Minimum Functionality test (MFT), is a collection of simple examples (and labels) to check a
behavior within a capability. MFTs are similar to
creating small and focused testing datasets, and are
particularly useful for detecting when models use
shortcuts to handle complex inputs without actually
mastering the capability.
...An Invariance test (INV) is when we apply
label-preserving perturbations to inputs and expect
the model prediction to remain the same.
A Directional Expectation test (DIR) is similar,
except that the label is expected to change in a certain way. For example, we expect that sentiment
will not become more positive if we add “You are
lame.” to the end of tweets directed at an airline
(Figure 1C).
I haven't been actively involved in NLG for long, so this answer will be a bit more anecdotal than SO's algorithms would like. Starting with the fact that in my corner of Europe, the general sentiment toward peer review requirements for any kind of NLG project are higher by several orders of magnitude compared to other sciences - and likely not without reason or tensor thereof.
This makes funding a bigger challenge, so wherever you are, I wish you luck on that front. I'm not sure of how big of a deal this site is in the niche, but [Ehud Reiter's Blog][1] is where I would start looking into your tooling ideas.
Maybe even reach out to them/him personally, because I can't think of another source that has an academic background and a strong propensity for practical applications of NLG, at least based on the kind of content they've been putting out over the years.
Your background, environment/funding, and seniority level/control you have over the project will eventually compose your vector decision for you. I's just how it goes on the bleeding edge of anything. What I will add, though, is not to limit yourself to a single language or technology in this phase because of those precise reasons you've mentioned. I'd recommend the same in terms of potential open source involvement but if your profile information is accurate, that probably won't happen, no matter what you do and accomplish.
But yeah, in the grand scheme of things, your question is far from too broad, in my view. It identifies a rather unmistakable problem pattern that not all branches of science are as lackadaisical to approach as NLG-adjacent fields seem to be right now. In that regard, it's not broad enough and will need to be promulgated far and wide before community-driven tooling will give you serious options on a micro level.
Blasphemy, sure, but the performance is already stacked against you As for the question potentially being too broad, I'd posit it is not broad enough, so long as we collectively remain in a "oh, I was waiting for you to start doing something about it" phase.
P.S. I'd eliminate any Rust and ECMAScript alternatives prior to looking into Python, blapshemous as this might sound to a 2021 data scientist
. Some might ARight nowccounting forr the ridicule this would receive xou sltrsfx hsbr s fszs drz zhsz s mrnzsl rcrtvidr, sz lrsdz
due to performance easons.
[1]: https://ehudreiter.com/2016/12/18/nlg-vs-templates/
I'm embarking on a project for a non-profit organization to help process and classify 1000's of reports annually from their field workers / contractors the world over. I'm relatively new to NLP and as such wanted to seek the group's guidance on the approach to solve our problem.
I'll highlight the current process, and our challenges and would love your help on the best way to solve our problem.
Current process: Field officers submit reports from locally run projects in the form of best practices. These reports are then processed by a full-time team of curators who (i) ensure they adhere to a best-practice template and (ii) edit the documents to improve language/style/grammar.
Challenge: As the number of field workers increased the volume of reports being generated has grown and our editors are now becoming the bottle-neck.
Solution: We would like to automate the 1st step of our process i.e., checking the document for compliance to the organizational best practice template
Basically, we need to ensure every report has 3 components namely:
1. States its purpose: What topic / problem does this best practice address?
2. Identifies Audience: Who is this for?
3. Highlights Relevance: What can the reader do after reading it?
Here's an example of a good report submission.
"This document introduces techniques for successfully applying best practices across developing countries. This study is intended to help low-income farmers identify a set of best practices for pricing agricultural products in places where there is no price transparency. By implementing these processes, farmers will be able to get better prices for their produce and raise their household incomes."
As of now, our approach has been to use RegEx and check for keywords. i.e., to check for compliance we use the following logic:
1 To check "states purpose" = we do a regex to match 'purpose', 'intent'
2 To check "identifies audience" = we do a regex to match with 'identifies', 'is for'
3 To check "highlights relevance" = we do a regex to match with 'able to', 'allows', 'enables'
The current approach of RegEx seems very primitive and limited so I wanted to ask the community if there is a better way to solving this problem using something like NLTK, CoreNLP.
Thanks in advance.
Interesting problem, i believe its a thorough research problem! In natural language processing, there are few techniques that learn and extract template from text and then can use them as gold annotation to identify whether a document follows the template structure. Researchers used this kind of system for automatic question answering (extract templates from question and then answer them). But in your case its more difficult as you need to learn the structure from a report. In the light of Natural Language Processing, this is more hard to address your problem (no simple NLP task matches with your problem definition) and you may not need any fancy model (complex) to resolve your problem.
You can start by simple document matching and computing a similarity score. If you have large collection of positive examples (well formatted and specified reports), you can construct a dictionary based on tf-idf weights. Then you can check the presence of the dictionary tokens. You can also think of this problem as a binary classification problem. There are good machine learning classifiers such as svm, logistic regression which works good for text data. You can use python and scikit-learn to build programs quickly and they are pretty easy to use. For text pre-processing, you can use NLTK.
Since the reports will be generated by field workers and there are few questions that will be answered by the reports (you mentioned about 3 specific components), i guess simple keyword matching techniques will be a good start for your research. You can gradually move to different directions based on your observations.
This seems like a perfect scenario to apply some machine learning to your process.
First of all, the data annotation problem is covered. This is usually the most annoying problem. Thankfully, you can rely on the curators. The curators can mark the specific sentences that specify: audience, relevance, purpose.
Train some models to identify these types of clauses. If all the classifiers fire for a certain document, it means that the document is properly formatted.
If errors are encountered, make sure to retrain the models with the specific examples.
If you don't provide yourself hints about the format of the document this is an open problem.
What you can do thought, is ask people writing report to conform to some format for the document like having 3 parts each of which have a pre-defined title like so
1. Purpose
Explains the purpose of the document in several paragraph.
2. Topic / Problem
This address the foobar problem also known as lorem ipsum feeling text.
3. Take away
What can the reader do after reading it?
You parse this document from .doc format for instance and extract the three parts. Then you can go through spell checking, grammar and text complexity algorithm. And finally you can extract for instance Named Entities (cf. Named Entity Recognition) and low TF-IDF words.
I've been trying to do something very similar with clinical trials, where most of the data is again written in natural language.
If you do not care about past data, and have control over what the field officers write, maybe you can have them provide these 3 extra fields in their reports, and you would be done.
Otherwise; CoreNLP and OpenNLP, the libraries that I'm most familiar with, have some tools that can help you with part of the task. For example; if your Regex pattern matches a word that starts with the prefix "inten", the actual word could be "intention", "intended", "intent", "intentionally" etc., and you wouldn't necessarily know if the word is a verb, a noun, an adjective or an adverb. POS taggers and the parsers in these libraries would be able to tell you the type (POS) of the word and maybe you only care about the verbs that start with "inten", or more strictly, the verbs spoken by the 3rd person singular.
CoreNLP has another tool called OpenIE, which attempts to extract relations in a sentence. For example, given the following sentence
Born in a small town, she took the midnight train going anywhere
CoreNLP can extract the triple
she, took, midnight train
Combined with the POS tagger for example; you would also know that "she" is a personal pronoun and "took" is a past tense verb.
These libraries can accomplish many other tasks such as tokenization, sentence splitting, and named entity recognition and it would be up to you to combine all of these tools with your domain knowledge and creativity to come up with a solution that works for your case.
I was wondering if there's a NLP/ML technique for this.
Suppose given a set of sentences,
I watched the movie.
Heard the movie is great, have to watch it.
Got the tickets for the movie.
I am at the movie.
If i have to assign a probability to each of these sentences, that they have "actually" watched the movie, i would assign it in decreasing order of 1,4,3,2.
Is there a way to do this automatically, using some classifier or rules? Any paper/link would help.
These are common issues in textual entailment. I'll refer you to some papers. While their motivation is for textual entailment, I believe your problem should be easier than that.
Determining Modality and Factuality for Textual Entailment
Learning to recognize features of valid textual entailments
Some of these suggestions should help you decide on some features/keywords to consider when ranking.
Except 1, none of the other statements necessarily imply that the person has watched the movie. They could have bought the tickets for somebody else (3) and might be the person who sells popcorn outside the halls (4). I don't think there is any clever system out there that will read between the lines for each sentence and return an answer that exactly agrees with your intuitions (which might be different from that of other people for the same sentence btw).
If this strangely is the only case that you care about (which is possible if you are explicitly working with movie reviews), then it might be worth your time to come with a large number of heuristics patched together that yields a function that near exactly agrees with your intuitions about this.
Otherwise look for context available in all the other sentences these sentences originate from to find relevant clues. Somebody who has actually watched the movie may comment on how they liked it, express opinions about specific scenes, characters and actors from the movie, etc. So if the text contains a lot of high sentiment sentences and refers to words and phrases from the movie, then the person has probably watched the movie. If a lot of it is in future tense, then maybe not.
If you are working with an specific domain, such as "watched the movie or not", or maybe more generally "attended to an event or not", it's basically a case of the Text Classification task.
The common approach in NLP is to use a large amount of sentences tagged as watched or didn't watch to train a machine learning based classifier. The most commonly used features are the presence/absence of keywords, bigrams (sequences of 2 words) and maybe trigrams (sequences of 3 words).
Since you talked about probability, things may get a little more complex. As adi92 noted, in 3 of your sentences the answer is not clear. A way to represent that in the training data could be that a sentence with 0.3 probability of watched appear 3 times tagged as watched and 7 as didn't watch. Most classifiers can have their output easily turned into probabilities.
Anyway, I believe that the main difficulty would be creating a training dataset for the task.
I have no clue of where to start on this. I've never done any NLP and only programmed in Python 3.1, which I have to use. I'm looking at the site http://www.linkedin.com and I have to gather all of the public profiles and some of them have very fake names, like 'aaaaaa k dudujjek' and I've been told I can use NLP to find the real names, where would I even start?
This is a difficult problem to solve, and one which starts with acquiring valid given name & surname lists.
How large is the set of names that you're evaluating, and where do they come from? These are both important things for you to consider. If you're evaluating a small set of "American" names, your valid name lists will differ greatly from lists of Japanese or Indian names, for instance.
Your idea of scraping LinkedIn is on the right track, but you were right to catch the fake profile/name flaw. A better website would probably be something like IMDB (perhaps scraping names by iterating over different birth years), or Wikipedia's lists of most popular given names and most common surnames.
When it comes down to it, this is a precision vs. recall problem: in order to miss fewer fakes, you're inevitably going to throw out some real names. If you loosen up your restrictions, you'll get more fakes, but you'll also throw out fewer real names.
Several possibilities here, but the most obvious seems to be with HMMs, i.e. Hidden Markov Models. The NLTK kit includes [at least] one module for HMMs, although I must admit I never used it.
Another possible snag is that AFAIK, NTLK is not yet ported to Python 3.0
This said, and while I'm quite keen on using NLP techniques where applicable, I think that a process which would use several paradigms, including some NLP tricks may be a better solution for this particular problem. For example, storing even a reduced dictionary of common family names (and first names) in a traditional database may offer both a more reliable and more computationally efficient way of filtering a significant portion of the input data, leaving precious CPU resources to be spent on less obvious cases.
i am afraid this problem is not solveable if your list is even only minimally ‘open’ — if the names are eg customers from a small traditionally acting population, you might end up with a few hundred names for thousands of people. but generally you can hardly predict what is a real name and what is not, however unusual an arabic, chinese, or bantu name may look in a sample of, say, south english rural neighborhood names. i mean, ‘Ng’ is a common cantonese surname, and ‘O’ is common in korea, so assumptions may fail. there is this place in austria called ‘fucking’, so even looking out for four letter words is no guarantee for success.
what you could do is work through a sufficiently big sample of such names and sort them out manually. then, use all kinds of textprocessing tools and collect metrics. maybe you can derive a certain likelyhood for a name to be recognized as fake, maybe it will not be viable. you will never go beyond likelyhoods here, though.
as an aside, we used to use google maps and the telephone directory for validating customer data years ago. if google maps could find the place, we called the address validated. it is clear that under stricter requirements, true validation must go much further. let’s not forget the validation of such data is much more a social question than a linguistic one.
I want to analyze answers to a web survey (Git User's Survey 2008 if one is interested). Some of the questions were free-form questions, like "How did you hear about Git?". With more than 3,000 replies analyzing those replies entirely by hand is out of the question (especially that there is quite a bit of free-form questions in this survey).
How can I group those replies (probably based on the key words used in response) into categories at least semi-automatically (i.e. program can ask for confirmation), and later how to tabularize (count number of entries in each category) those free-form replies (answers)? One answer can belong to more than one category, although for simplicity one can assume that categories are orthogonal / exclusive.
What I'd like to know is at least keyword to search for, or an algorithm (a method) to use. I would prefer solutions in Perl (or C).
Possible solution No 1. (partial): Bayesian categorization
(added 2009-05-21)
One solution I thought about would be to use something like algorithm (and mathematical method behind it) for Bayesian spam filtering, only instead of one or two categories ("spam" and "ham") there would be more; and categories itself would be created adaptively / interactively.
Text::Ngrams + Algorithm::Cluster
Generate some vector representation for each answer (e.g. word count) using Text::Ngrams.
Cluster the vectors using Algorithm::Cluster to determine the groupings and also the keywords which correspond to the groups.
You are not going to like this. But: If you do a survey and you include lots of free-form questions, you better be prepared to categorize them manually. If that is out of the question, why did you have those questions in the first place?
I've brute forced stuff like this in the past with quite large corpuses. Lingua::EN::Tagger, Lingua::Stem::En. Also the Net::Calais API is (unfortunately, as Thomposon Reuters are not exactly open source friendly) pretty useful for extracting named entities from text. Of course once you've cleaned up the raw data with this stuff, the actual data munging is up to you. I'd be inclined to suspect that frequency counts and a bit of mechanical turk cross-validation of the output would be sufficient for your needs.
Look for common words as keywords, but through away meaningless ones like "the", "a", etc. After that you get into natural language stuff that is beyond me.
It just dawned on me that the perfect solution for this is AAI (Artificial Artificial Intelligence). Use Amazon's Mechanical Turk. The Perl bindings are Net::Amazon::MechanicalTurk. At one penny per reply with a decent overlap (say three humans per reply) that would come to about $90 USD.