I'm making a simple search engine, and as I go through the documents that are going to be indexed, I want to automatically identify the words that should be ignored (such as "and" and "the").
The only simple method I can think of is just ignore words of up to a certain length (if they're not lengthy enough, then they're considered stop words). Any other method would probably have to require data mining (I'm open to suggestions).
I would prefer a method that I can use as i go through the documents, but I'm open to the other suggestions. I just need a simple method.
Short answer is: don't. As in don't bother, but instead strip them from the query and/or weigh them appropriately by TF-IDF.
Quoting the Xapian manual: http://xapian.org/docs/stemming.html
It has been traditional in setting up IR systems to discard the very commonest words of a language - the stopwords - during indexing. A more modern approach is to index everything, which greatly assists searching for phrases for example. Stopwords can then still be eliminated from the query as an optional style of retrieval. In either case, a list of stopwords for a language is useful.
Getting a list of stopwords can be done by sorting a vocabulary of a text corpus for a language by frequency, and going down the list picking off words to be discarded.
Related
For those who are not familiar with what a homophone is, I provide the following examples:
our & are
hi & high
to & too & two
While using the Speech API included with iOS, I am encountering situations where a user may say one of these words, but it will not always return the word I want.
I looked into the [alternativeSubstrings] (link) property wondering if this would help, but in my testing of the above words, it always comes back empty.
I also looked into the Natural Language API, but could not find anything in there that looked useful.
I understand that as a user adds more words, the Speech API can begin to infer context and correct for these, but my use case will not work well with this since it will often only want one or two words at most, limiting the effectiveness of context.
An example of contextual processing:
Using the words above on their own, I get these results:
are
hi
to
However, if I put together the following sentence, you can see they are all wrong:
I am too high for our ladder
Ideally, I would either get a list back containing [are, our], [to, too, two], [hi, high] for each transcription segment, or would have a way to compare a string against a function that supports homophones.
An example of this would be:
if myDetectedWord == "to" then { ... }
Where myDetectedWord can be [to, too, two], and this function would return true for each of these.
This is a common NLP dilemma, and I'm not so sure what might be your desired output in this application. However, you may want to bypass this problem in your design/architecture process, if possible and if you could. Otherwise, this problem is to turn into a challenge.
Being said that, if you wish to really get into it, I like this idea of yours:
string against a function
This might be more efficient and performance friendly.
One way, I'd be liking to solve this problem would be though RegEx processing, instead of using endless loops and arrays. You could maybe prototype loops and arrays to begin with and see how it works, then you might want to use regular expression for gaining performance.
You could for instance define fixed arrays in regular expressions and quickly check against your string (word by word, maybe using back-referencing) and you can add many boundaries in your expressions for string processing, as you wish.
Your fixed arrays also can be designed based on probabilities of occurring certain words in certain part of a string. For instance,
^I
vs
^eye
The probability of I being the first word is much higher than that of eye.
The probability of I in any part of a string is higher than that of eye, also.
You might want to weight words based on that.
I'd say the key would be that you'd narrow down your desired outputs as focused as possible and increase accuracy, [maybe even with 100 words if possible], if you wish to have a good/working application.
Good project though, I hope you like/enjoy the challenge.
I am working on developing a tool for language identification of a given text i.e. given a sample text, identify the language (for e.g. English, Swedish, German, etc.) it is written in.
Now the strategy I have decided to follow (based on a few references I have gathered) are as follows -
a) Create a character n-gram model (The value of n is decided based on certain heuristics and computations)
b) Use a machine learning classifier(such as naive bayes) to predict the language of the given text.
Now, the doubt I have is - Is creating a character N-gram model necessary. As in, what disadvantage does a simple bag of words strategy have i.e. if I use all the words possible in the respective language to create a prediction model, what could be the possible cases where it would fail.
The reason why this doubt arose was the fact that any reference document/research paper I've come across states that language identification is a very difficult task. However, just using this strategy of using the words in the language seems to be a simple task.
EDIT: One reason why N-gram should be preferred is to make the model robust even if there are typos as stated here. Can anyone point out more?
if I use all the words possible in the respective language to create a prediction model, what could be the possible cases where it would fail
Pretty much the same cases were a character n-gram model would fail. The problem is that you're not going to find appropriate statistics for all possible words.(*) Character n-gram statistics are easier to accumulate and more robust, even for text without typos: words in a language tend to follow the same spelling patterns. E.g. had you not found statistics for the Dutch word "uitbuiken" (a pretty rare word), then the occurrence of the n-grams "uit", "bui" and "uik" would still be strong indicators of this being Dutch.
(*) In agglutinative languages such as Turkish, new words can be formed by stringing morphemes together and the number of possible words is immense. Check the first few chapters of Jurafsky and Martin, or any undergraduate linguistics text, for interesting discussions on the possible number of words per language.
Cavnar and Trenkle proposed a very simple yet efficient approach using character n-grams of variable length. Maybe you should try to implement it first and move to a more complex ML approach if C&T approach doesn't meet your requirements.
Basically, the idea is to build a language model using only the X (e.g. X = 300) most frequent n-grams of variable length (e.g. 1 <= N <= 5). Doing so, you are very likely to capture most functional words/morphemes of the considered language... without any prior linguistic knowledge on that language!
Why would you choose character n-grams over a BoW approach? I think the notion of character n-gram is pretty straightforward and apply to every written language. Word, is a much much complex notion which greatly differ from one language to another (consider languages with almost no spacing marks).
Reference: http://odur.let.rug.nl/~vannoord/TextCat/textcat.pdf
The performance really depends on your expected input. If you will be classifying multi-paragraph text all in one language, a functional words list (which your "bag of words" with pruning of hapaxes will quickly approximate) might well serve you perfectly, and could work better than n-grams.
There is significant overlap between individual words -- "of" could be Dutch or English; "and" is very common in English but also means "duck" in the Scandinavian languages, etc. But given enough input data, overlaps for individual stop words will not confuse your algorithm very often.
My anecdotal evidence is from using libtextcat on the Reuters multilingual newswire corpus. Many of the telegrams contain a lot of proper names, loan words etc. which throw off the n-gram classifier a lot of the time; whereas just examining the stop words would (in my humble estimation) produce much more stable results.
On the other hand, if you need to identify short, telegraphic utterances which might not be in your dictionary, a dictionary-based approach is obviously flawed. Note that many North European languages have very productive word formation by free compounding -- you see words like "tandborstställbrist" and "yhdyssanatauti" being coined left and right (and Finnish has agglutination on top -- "yhdyssanataudittomienkinkohan") which simply cannot be expected to be in a dictionary until somebody decides to use them.
I am looking to write a basic profanity filter in a Rails based application. This will use a simply search and replace mechanism whenever the appropriate attribute gets submitted by a user. My question is, for those who have written these before, is there a CSV file or some database out there where a list of profanity words can be imported into my database? We are submitting the words that we will replace the profanities with on our own. We more or less need a database of profanities, racial slurs and anything that's not exactly rated PG-13 to get triggered.
As the Tin Man suggested, this problem is difficult, but it isn't impossible. I've built a commercial profanity filter named CleanSpeak that handles everything mentioned above (leet speak, phonetics, language rules, whitelisting, etc). CleanSpeak is capable of filtering 20,000 messages per second on a low end server, so it is possible to build something that works well and performs well. I will mention that CleanSpeak is the result of about 3 years of on-going development though.
There are a few things I tell everyone that is looking to try and tackle a language filter.
Don't use regular expressions unless you have a small list and don't mind a lot of things getting through. Regular expressions are relatively slow overall and hard to manage.
Determine if you want to handle conjugations, inflections and other language rules. These often add a considerable amount of time to the project.
Decide what type of performance you need and whether or not you can make multiple passes on the String. The more passes you make the slow your filter will be.
Understand the scunthrope and clbuttic problems and determine how you will handle these. This usually requires some form of language intelligence and whitelisting.
Realize that whitespace has a different meaning now. You can't use it as a word delimiter any more (b e c a u s e of this)
Be careful with your handling of punctuation because it can be used to get around the filter (l.i.k.e th---is)
Understand how people use ascii art and unicode to replace characters (/ = v - those are slashes). There are a lot of unicode characters that look like English characters and you will want to handle those appropriately.
Understand that people make up new profanity all the time by smashing words together (likethis) and figure out if you want to handle that.
You can search around StackOverflow for my comments on other threads as I might have more information on those threads that I've forgotten here.
Here's one you could use: Offensive/Profane Word List from CMU site
Based on personal experience, you do understand that it's an exercise in futility?
If someone wants to inject profanity, there's a slew of words that are innocent in one context, and profane in another so you'll have to write a context parser to avoid black-listing clean words. A quick glance at CMU's list shows words I'd never consider rude/crude/socially unacceptable. You'll see there are many words that could be proper names or nouns, countries, terms of endearment, etc. And, there are myriads of ways to throw your algorithm off using L33T speak and such. Search Wikipedia and the internets and you can build tables of variations of letters.
Look at CMU's list and imagine how long the list would be if, in addition to the correct letter, every a could also be 4, o could be 0 or p, e could be 3, s could be 5. And, that's a very, very, short example.
I was asked to do a similar task and wrote code to generate L33T variations of the words, and generated a hit-list of words based on several profanity/offensive lists available on the internet. After running the generator, and being a little over 1/4 of the way through the file, I had over one million entries in my DB. I pulled the plug on the project at that point, because the time spent searching, even using Perl's Regex::Assemble, was going to be ridiculous, especially since it'd still be so easy to fool.
I recommend you have a long talk with whoever requested that, and ask if they understand the programming issues involved, and low-likelihood of accuracy and success, especially over the long-term, or the possible customer backlash when they realize you're censoring them.
I have one that I've added to (obfuscated a bit) but here it is: https://github.com/rdp/sensible-cinema/blob/master/lib/subtitle_profanity_finder.rb
I'm making an application using a dependency tree parser. Actually, the parser is this one:
Parser Stanford, but it rarely change one or two letters of some words in a sentence that I want to parse. This is a big trouble for me, because I can't see any pattern in these changes and I need the dependency tree with the same words of my sentence.
All I can see is that just some words have these problems. I'm working with a tweets database. So, I have a lot of grammar mistakes in this data. For example the hashtag '#AllAmericanhumour ' becomes AllAmericanhumor. It misses one letter(u).
Is there anything I can do to solve this problem? In my first view I thought using an edit distance algorithm, but I think that might be an easier way to do it.
Thanks everybody in advance
You can give options to the tokenizer with the -tokenize.options flag/property. For this particular normalization, you can turn it off with
-tokenize.options americanize=false
There are also various other normalizations that you can turn off (see PTBTokenizer or http://nlp.stanford.edu/software/tokenizer.shtml. You can turn off a lot with
-tokenize.options ptb3Escaping=false
However, the parser is trained on data that looks like the output of ptb3Escaping=true and so will tend to degrade in performance if used with unnormalized tokens. So, you may want to consider alternative strategies.
If you're working at the Java level, you can look at the word tokens, which are actually Maps, and they have various keys. OriginalTextAnnotation will give you the unnormalized token, even when it has been normalized. CharacterOffsetBeginAnnotation and CharacterOffsetEndAnnotation will map to character offsets into the text.
p.s. And you should accept some answers :-).
I've done some Google searching but couldn't find what I was looking for.
I'm developing a scrabble-type word game in rails, and was wondering if there was a simple way to validate what the player inputs in the game is actually a word. They'd be typing the word out.
Is validation against some sort of English language dictionary database loaded within the app best way to solve this problem? If so, are there any libraries that offer this kind of functionality? If not, what would you suggest?
Thanks for your help!
You need two things:
a word list
some code
The word list is the tricky part. On most Unix systems there's a word list at /usr/share/dict/words or /usr/dict/words -- see http://en.wikipedia.org/wiki/Words_(Unix) for more details. The one on my Mac has 234,936 words in it. But they're not all valid Scrabble words. So you'd have to somehow acquire a Scrabble dictionary, make sure you have the right license to use it, and process it so it's a text file.
(Update: The word list for LetterPress is now open source, and available on GitHub.)
The code is no problem in the simple case. Here's a script I whipped up just now:
words = {}
File.open("/usr/share/dict/words") do |file|
file.each do |line|
words[line.strip] = true
end
end
p words["magic"]
p words["saldkaj"]
This will output
true
nil
I leave it as an exercise for the reader to make it into a proper Words object. (Technically it's not a Dictionary since it has no definitions.) Or to use a DAWG instead of a hash, even though a hash is probably fine for your needs.
A piece of language-agnostic advice here, is that if you only care about the existence of a word (which in such a case, you do), and you are planning to load the entire database into the application (which your query suggests you're considering) then a DAWG will enable you to check the existence in O(n) time complexity where n is the size of the word (dictionary size has no effect - overall the lookup is essentially O(1)), while being a relatively minimal structure in terms of memory (indeed, some insertions will actually reduce the size of the structure, a DAWG for "top, tap, taps, tops" has fewer nodes than one for "tops, tap").