It's difficult to tell what is being asked here. This question is ambiguous, vague, incomplete, overly broad, or rhetorical and cannot be reasonably answered in its current form. For help clarifying this question so that it can be reopened, visit the help center.
Closed 11 years ago.
I need to write a module to detect similar documents. I have read many papers of fingerprints of documents techniques and others, but I do not know how to write code or implement such a solution. The algorithm should work for Chinese, Japanese, English and German language or be language independent. How can I accomplish this?
Bayesian filters have exactly this purpose. That's the techno you'll find in most tools that identify spam.
Example, to detect a language (from http://sebsauvage.net/python/snyppets/#bayesian) :
from reverend.thomas import Bayes
guesser = Bayes()
guesser.train('french','La souris est rentrée dans son trou.')
guesser.train('english','my tailor is rich.')
guesser.train('french','Je ne sais pas si je viendrai demain.')
guesser.train('english','I do not plan to update my website soon.')
>>> print guesser.guess('Jumping out of cliffs it not a good idea.')
[('english', 0.99990000000000001), ('french', 9.9999999999988987e-005)]
>>> print guesser.guess('Demain il fera très probablement chaud.')
[('french', 0.99990000000000001), ('english', 9.9999999999988987e-005)]
But it works to detect any type you will train it for : technical text, songs, jokes, etc. As long as you can provide enought material to let the tool learn what does you document looks like.
If these are pure text documents, or you have a method to extract the text from the documents, you can use a technique called shingling.
You first compute a unique hash for each document. If these are the same, you are done.
If not, you break each document down into smaller chunks. These are your 'shingles.'
Once you have the shingles, you can then compute identity hashes for each shingle and compare the hashes of the shingles to determine if the documents are actually the same.
The other technique you can use is to generate n-grams of the entire documents and compute the number of similar n-grams in each document and produce a weighted score for each document. Basically an n-gram is splitting a word into smaller chunks. 'apple' would become ' a', ' ap', 'app', 'ppl', 'ple', 'le '. (This is technically a 3-gram) This approach can become quite computationally expensive over a large number of documents or over two very large documents. Of course, common n-grams 'the', ' th, 'th ', etc need to be weighted to score them lower.
I've posted about this on my blog and there are some links in the post to a few other articles on the subject Shingling - it's not just for roofers.
Best of luck!
Similarity can be found easily without classification. Try this O(n2) but works fine.
def jaccard_similarity(doc1, doc2):
a = sets(doc1.split())
b = sets(doc2.split())
similarity = float(len(a.intersection(b))*1.0/len(a.union(b))) #similarity belongs to [0,1] 1 means its exact replica.
return similarity
You can use or at last study difflib from Python's stdlib to write your code.
It is very flexible, and has algorithms to find differences between lists of strings, and to point these differences. Then you can use the get_close_matches() to find similar words:
>>> get_close_matches('appel', ['ape', 'apple', 'peach', 'puppy'])
['apple', 'ape']
It is not the solution but maybe it is a start.
You need to make your question more concrete. If you've already read the fingerprinting papers, you already know the principles at work, so describing common approaches here would not be beneficial. If you haven't, you should also check out papers on "duplicate detection" and various web spam detection related papers that have come out of Stanford, Google, Yahoo, and MS in recent years.
Are you having specific problems with coding the described algorithms?
Trouble getting started?
The first thing I'd probably do is separate the tokenization (the process of extracting "words" or other sensible sequences) from the duplicate detection logic, so that it is easy to plug in different parsers for different languages and keep the duplicate detection piece the same.
There is a rather good talk on neural networks on Google Techtalks that talks about using layered Boltzmann machines to generate feature vectors for documents that can then be used to measure document distance. The main issue is the requirement to have a large sample document set to train the network to discover relevant features.
If you're prepared to index the files that you want to search amongst, Xapian is an excellent engine, and provides Python bindings:
http://xapian.org/
http://xapian.org/docs/bindings/python/
If you are trying to detect the documents that are talking about the same topic, you could try collecting the most frequently used words, throw away the stop words . Documents that have a similar distribution of the most frequently used words are probably talking about similar things. You may need to do some stemming and extend the concept to n-grams if you want higher accuracy. For more advanced techniques, look into machine learning.
I think Jeremy has hit the nail on the head - if you just want to detect if files are different, a hash algorithm like MD5 or SHA1 is a good way to go.
Linus Torvalds' Git source control software uses SHA1 hashing in just this way - to check when files have been modified.
You might want to look into the DustBuster algorithm as outlined in this paper.
From the paper, they're able to detect duplicate pages without even examining the page contents. Of course examining the contents increases the efficacy, but using raw server logs is adequate for the method to detect duplicate pages.
Similar to the recommendation of using MD5 or SHA1 hashes, the DustBuster method largely relies on comparing file size as it primary signal. As simple as it sounds, it's rather effective for an initial first pass.
Related
I'm trying to write a python program that will decide if a given post is about the topic of volunteering. My data-sets are small (only the posts, which are examined 1 by 1) so approaches like LDA do not yield results.
My end goal is a simple True/False, a post is about the topic or not.
I'm trying this approach:
Using Google's word2vec model, I'm creating a "cluster" of words that are similar to the word: "volunteer".
CLUSTER = [x[0] for x in MODEL.most_similar_cosmul("volunteer", topn=120)]
Getting the posts and translating them to English, using Google translate.
Cleaning the translated posts using NLTK (removing stopwords, punctuation, and lemmatize the post)
Making a BOW out of the translated, clean post.
This stage is difficult for me. I want to calculate a "distance" / "similarity" / something that will help me get the True/False answer that I'm looking for, but I can't think of a good way to do that.
Thank you for your suggestions and help in advance.
You are attempting to intuitively improvise a set of steps that, in the end, will classify these posts into the two categories, "volunteering" and "not-volunteering".
You should looks for online examples that do "text classification" that are similar to your task, work through them (with their original demo data) for understanding, then adapt them incrementally to work with your data instead.
At some point, word2vec might be a helpful contributor to your task - but I wouldn't start with it. Similarly, eliminating stop-words, performing lemmatization, etc might eventually be helpful, but need not be important up front.
You'll typically want to start by acquiring (by hand-labeling if necessary) a training set of text for which you know the "volunteering" or "not-volunteering" value (known labels).
Then, create some feature-vectors for the texts – A simple starting approach that offers a quick baseline for later improvements is a "bag of words" representation.
Then, feed those representations, with the known-labels, to some existing classification algorithm. The popular scikit-learn package in Python offers many. That is: you don't yet need to be worrying about choosing ways to calculate a "distance" / "similarity" / something that will guide your own ad hoc classifier. Just feed the labeled data into one (or many) existing classifiers, and check how well they're doing. Many will be using various kinds of similarity/distance calculations internally - but that's automatic and explicit from choosing & configuring the algorithm.
Finally, when you have something working start-to-finish, no matter how modest in results, then try alternate ways of preprocessing text (stop-word-removal, lemmatization, etc), featurizing text, and alternate classifiers/algorithm paramterizations - to compare results, and thus discover what works well given your specific data, goals, and practical constraints.
The scikit-learn "Working With Text Data" guide is worth reviewing & working-through, and their "Choosing the right estimator" map is useful for understanding the broad terrain of alternate techniques and major algorithms, and when different ones apply to your task.
Also, scikit-learn contributors/educators like Jake Vanderplas (github.com/jakevdp) and Olivier Grisel (github.com/ogrisel) have many online notebooks/tutorials/archived-video-presentations which step through all the basics, often including text-classification problems much like yours.
Apologies for the vague nature of this question but I'm honestly not quite sure where to start and thought I'd ask here for guidance.
As an exercise, I've downloaded several academic papers and stored them as plain text in a mongoDB database.
I'd like to do write a search feature (using Python, R, whatever) that when you enter text and returns the most relevant articles. Clearly, relevant is really hard -- that's what google got so right.
However, I'm not looking for it to be perfect. Just to get something. A few thoughts I had were:
1) Simple MongoDB full text search
2) Implement Lucene Search
3) Tag them (unsure how though) and then return them sorted by the most number of tags?
Is there a solution someone has used that's out of the box and works fairly well? I can always optimize the search feature later -- for now I just want all the pieces to move together...
Thanks!
Is there a solution someone has used that's out of the box and works fairly well?
It depends on how you define well, but in simple terms, I'd say no. There is just no single and accurate definition of fairly well. A lot of challenges intrinsic to a particular problem arise when one trying to implement a good search algorithm. Those challenges lies in:
users needs diversity. Users in different fields have different intentions and as a result different expectation from a search result page;
natural languages diversity, if you are trying to implement multi-language search (German has a lot of noun compounds, Russian has enormous flexion variability etc.);
There are some algorithms that are proven to work better than others though, thus are good to start from. TF*IDF and BM25 two most popular.
I can always optimize the search feature later -- for now I just want all the pieces to move together...
MongoDB or any RDBMS with fulltext indexing support is good enough for a proof-of-concept, but if you need to optimize for search performance, you will need an inverted index (Solr/Lucene). From Solr/Lucene you will get ability to manage:
how exactly words are stemmed (this is important to solve undersemming/overstemming problems);
what the word is. Is "supercomputer" one word? What about "stackoverflow" or "OutOfBoundsException"?
synonyms and word expansion (should "O2" be found for a "oxygen" query?)
how exactly search is performed. Which words could be ignored during search. Which ones are required to be found. Which one are required to be found near each other (think of search phrase: "not annealed" or "without expansion").
This is just what comes to mind first.
So if you are planning to work these things out I definitely recommend Lucene as a framework or Solr/ElasticSearch as a search system if you need to build proof-of-concept fast. If not, MongoDB/RDMS will work well.
The context is : I already have clusters of words (phrases actually) resulting from kmeans applied to internet search queries and using common urls in the results of the search engine as a distance (co-occurrence of urls rather than words if I simplify a lot).
I would like to automatically label the clusters using semantics, in other words I'd like to extract the main concept surrounding a group of phrases considered together.
For example - sorry for the subject of my example - if I have the following bunch of queries : ['my husband attacked me','he was arrested by the police','the trial is still going on','my husband can go to jail for harrassing me ?','free lawyer']
My study deals with domestic violence, but clearly this cluster is focused on the legal aspect of the problem so the label could be "legal" for example.
I am new to NPL but I have to precise that I don't want to extract words using POS tagging (or at least this is not the expected final outcome but maybe a necessary preliminary step).
I read about Wordnet for sense desambiguation and I think that might be a good track, but I don't want to calculate similarity between two queries (since the clusters are the input) nor obtain the definition of one selected word thanks to the context provided by the whole bunch of words (which word to select in this case ?). I want to use the whole bunch of words to provide a context (maybe using synsets or categorization with the xml structure of the wordnet) and then summarize the context in one or few words.
Any ideas ? I can use R or python, I read a little about nltk but I don't find a way to use it in my context.
Your best bet is probably is to label the clusters manually, especially if there are few of them. This a difficult problem even for humans to solve, because you might need a domain expert. Anyone claiming they could do that automatically and reliably (except in some very limited domains) is probably running a startup and trying to get your business.
Also, going through the clusters yourself will have benefits. 1) you may discover you had the wrong number of clusters (k parameter) or that there was too much junk in the input to begin with. 2) you will gain qualitative insight into what is being talked about and what topic there are in the data (which you probably can't know before looking at the data). Therefore, label manually if qualitative insight is what you are after. If you need quantitative result too, you could then train a classifier on the manually labelled topics to 1) predict topics for the rest of the clusters, or 2) for future use, if you repeat the clustering, get new data, ...
When we talk about semantics in this area we mean Statistical Semantics. The statistical or distributional semantics is very different from other definitions of semantics which has logic and reasoning behind it. Statistical semantics is based on Distributional Hypothesis, which considers context as meaning aspect of words and phrases. Meaning in very abstract and general sense in different litterers is called topics. There are several unsupervised methods for modelling topics, such as LDA or even word2vec, which basically provide word similarity metric or suggest a list of similar words for a document as another context. Usually when you have these unsupervised clusters, you need a domain expert to tell the meaning of each cluster.
However, for several reasons you might accept low accuracy assignment of a word as the general topic (or as in your words "global semantic") to a list of phrases. If this is the case, I would suggest to take a look at Word Sense Disambiguation tasks which look for coarse grained word senses. For WordNet, it might be called supersense tagging task.
This paper worth to take a look: More or less supervised supersense tagging of Twitter
And about your question about choosing words from current phrases, there is also an active question about "converting phrase to vectors", my answer to that question in word2vec fashion might be useful:
How can a sentence or a document be converted to a vector?
I can add more related papers later if it comes to my mind.
The paper Automatic Labelling of Topic Models explains the author's approach to this problem. To provide an overview I can tell you that they generate some label candidates using the information retrieved from Wikipedia and Google, and once they have the list of candidates in place they rank those candidates to find the best label.
I think the code is not available online, but I have not looked for it.
The package chowmein claims to do this in python using the algorithm outlined in Automatic Labeling of Multinomial Topic Models.
One possible approach, which the below papers suggest is identifying the set of keywords from the cluster, getting all the synonyms and then finding the hypernyms for each synonym.
The idea is to get a more abstract meaning for the cluster by using the hypernym.
Example: A word cluster containing words dog and wolf should not be labelled with either word but as canids. They achieve it using synonymy and hypernymy.
Cluster Labeling by Word Embeddings
and WordNet’s Hypernymy
Automated Text Clustering and Labeling using Hypernyms
First: Any recs on how to modify the title?
I am using my own named entity recognition algorithm to parse data from plain text. Specifically, I am trying to extract lawyer practice areas. A common sentence structure that I see is:
1) Neil focuses his practice on employment, tax, and copyright litigation.
or
2) Neil focuses his practice on general corporate matters including securities, business organizations, contract preparation, and intellectual property protection.
My entity extraction is doing a good job of finding the key words, for example, my output from sentence one might look like this:
Neil focuses his practice on (employment), (tax), and (copyright litigation).
However, that doesn't really help me. What would be more helpful is if i got an output that looked more like this:
Neil focuses his practice on (employment - litigation), (tax - litigation), and (copyright litigation).
Is there a way to accomplish this goal using an existing python framework such as nltk (after my algo extracts the practice areas) can I use ntlk to extract the other words that my "practice areas" modify in order to get a more complete picture?
Named entity recognition (NER) systems typically use grammer-based rules or statistical language models. What you have described here seems to be based only on keywords, though.
Typically, and much like most complex NLP tasks, NER systems should be trained on domain-specific data so that they perform well on previously unseen (test) data. You will require adequate knowledge of machine learning to go down that path.
In "normal" language, if you want to extract words or phrases and categorize them into classes defined by you (e.g. litigation), if often makes sense to use category labels in external ontologies. An example could be:
You want to extract words and phrases related to sports.
Such a categorization (i.e. detecting whether or not a word is indeed related to sports) is not a "general"-enough problem. Which means you will not find ready-made systems that will solve the problem (e.g. algorithms in the NLTK library). You can, however, use an ontology like Wikipedia and exploit the category labels available there.
E.g., you can check that if you search Wikipedia for "football", which has a category label "ball games", which in turn is under "sports".
Note that the wikipedia category labels form a directed graph. If you build a system which exploits the category structure of such an ontology, you should be able to categorize terms in your texts as you see fit. Moreover, you can even control the granularity of the categorization (e.g. do you want just "sports", or "individual sports" and "team sports").
I have built such a system for categorizing terms related to computer science, and it worked remarkably well. The closest freely available system that works in a similar way is the Wikifier built by the cognitive computing group at the University of Illinois at Urbana-Champaign.
Caveat: You may need to tweak a simple category-based code to suit your needs. E.g. there is no wikipedia page for "litigation". Instead, it redirects you to a page titled "lawsuit". Such cases need to be handled separately.
Final Note: This solution is really not in the area of NLP, but my past experience suggests that for some domains, this kind of ontology-based approach works really well. Also, I have used the "sports" example in my answer because I know nothing about legal terminology. But I hope my example helps you understand the underlying process.
I do not think your "algo" is even doing entity recognition... however, stretching the problem you presented quite a bit, what you want to do looks like coreference resolution in coordinated structures containing ellipsis. Not easy at all: start by googling for some relevant literature in linguistics and computational linguistics. I use the standard terminology from the field below.
In practical terms, you could start by assigning the nearest antecedent (the most frequently used approach in English). Using your examples:
first extract all the "entities" in a sentence
from the entity list, identify antecedent candidates ("litigation", etc.). This is a very difficult task, involving many different problems... you might avoid it if you know in advance the "entities" that will be interesting for you.
finally, you assign (resolve) each anaphora/cataphora to the nearest antecedent.
Have a look at CogComp NER tagger:
https://github.com/CogComp/cogcomp-nlp/tree/master/ner
I realize this is a broad topic, but I'm looking for a good primer on parsing meaning from text, ideally in Python. As an example of what I'm looking to do, if a user makes a blog post like:
"Manny Ramirez makes his return for the Dodgers today against the Houston Astros",
what's a light-weight/ easy way of getting the nouns out of a sentence? To start, I think I'd limit it to proper nouns, but I wouldn't want to be limited to just that (and I don't want to rely on a simple regex that assumes anything Title Capped is a proper noun).
To make this question even worse, what are the things I'm not asking that I should be? Do I need a corpus of existing words to get started? What lexical analysis stuff do I need to know to make this work? I did come across one other question on the topic and I'm digging through those resources now.
You need to look at the Natural Language Toolkit, which is for exactly this sort of thing.
This section of the manual looks very relevant: Categorizing and Tagging Words - here's an extract:
>>> text = nltk.word_tokenize("And now for something completely different")
>>> nltk.pos_tag(text)
[('And', 'CC'), ('now', 'RB'), ('for', 'IN'), ('something', 'NN'),
('completely', 'RB'), ('different', 'JJ')]
Here we see that and is CC, a coordinating conjunction; now and completely are RB, or adverbs; for is IN, a preposition; something is NN, a noun; and different is JJ, an adjective.
Use the NLTK, in particular chapter 7 on Information Extraction.
You say you want to extract meaning, and there are modules for semantic analysis, but I think IE is all you need--and honestly one of the only areas of NLP computers can handle right now.
See sections 7.5 and 7.6 on the subtopics of Named Entity Recognition (to chunk and categorize Manny Ramerez as a person, Dodgers as a sports organization, and Houston Astros as another sports organization, or whatever suits your domain) and Relationship Extraction. There is a NER chunker that you can plugin once you have the NLTK installed. From their examples, extracting a geo-political entity (GPE) and a person:
>>> sent = nltk.corpus.treebank.tagged_sents()[22]
>>> print nltk.ne_chunk(sent)
(S
The/DT
(GPE U.S./NNP)
is/VBZ
one/CD
...
according/VBG
to/TO
(PERSON Brooke/NNP T./NNP Mossman/NNP)
...)
Note you'll still need to know tokenization and tagging, as discussed in earlier chapters, to get your text in the right format for these IE tasks.
Natural Language Processing (NLP) is the name for parsing, well, natural language. Many algorithms and heuristics exist, and it's an active field of research. Whatever algorithm you will code, it will need to be trained on a corpus. Just like a human: we learn a language by reading text written by other people (and/or by listening to sentences uttered by other people).
In practical terms, have a look at the Natural Language Toolkit. For a theoretical underpinning of whatever you are going to code, you may want to check out Foundations of Statistical Natural Language Processing by Chris Manning and Hinrich Schütze.
(source: stanford.edu)
Here is the book I stumbled upon recently: Natural Language Processing with Python
What you want is called NP (noun phrase) chunking, or extraction.
Some links here
As pointed out, this is very problem domain specific stuff. The more you can narrow it down, the more effective it will be. And you're going to have to train your program on your specific domain.
This is a really really complicated topic. Generally, this sort of stuff falls under the rubric of Natural Language Processing, and tends to be tricky at best. The difficulty of this sort of stuff is precisely why there still is no completely automated system for handling customer service and the like.
Generally, the approach to this stuff REALLY depends on precisely what your problem domain is. If you're able to winnow down the problem domain, you can gain some very serious benefits; to use your example, if you're able to determine that your problem domain is baseball, then that gives you a really strong head start. Even then, it's a LOT of work to get anything particularly useful going.
For what it's worth, yes, an existing corpus of words is going to be useful. More importantly, determining the functional complexity expected of the system is going to be critical; do you need to parse simple sentences, or is there a need for parsing complex behavior? Can you constrain the inputs to a relatively simple set?
Regular expressions can help in some scenario. Here is a detailed example: What’s the Most Mentioned Scanner on CNET Forum, which used a regular expression to find all mentioned scanners in CNET forum posts.
In the post, a regular expression as such was used:
(?i)((?:\w+\s\w+\s(?:(?:(?:[0-9]+[a-z\-]|[a-z]+[0-9\-]|[0-9])[a-z0-9\-]*)|all-in-one|all in one)\s(\w+\s){0,1}(?:scanner|photo scanner|flatbed scanner|adf scanner|scanning|document scanner|printer scanner|portable scanner|handheld scanner|printer\/scanner))|(?:(?:scanner|photo scanner|flatbed scanner|adf scanner|scanning|document scanner|printer scanner|portable scanner|handheld scanner|printer\/scanner)\s(\w+\s){1,2}(?:(?:(?:[0-9]+[a-z\-]|[a-z]+[0-9\-]|[0-9])[a-z0-9\-]*)|all-in-one|all in one)))
in order to match either of the following:
two words, then model number (including all-in-one), then “scanner”
“scanner”, then one or two words, then model number (including
all-in-one)
As a result, the text extracted from the post was like,
discontinued HP C9900A photo scanner
scanning his old x-rays
new Epson V700 scanner
HP ScanJet 4850 scanner
Epson Perfection 3170 scanner
This regular expression solution worked in a way.