Ask HN: How do you split strings (to get keywords)?

2 points by gopher 16 years ago | 7 comments

First trial, one splits on whitespace, but this sucks on interpunction and special characters.

Second trial, you use a alpha-numeric whitelist and split on anything else, but what about umlauts? What about hebrew or cyrillic?

Third trial: split on characters < 32, whitespace and interpunction characters; this works somehow but is ugly. What would you do to get keywords from a string?

TallGuyShort 16 years ago |

It depends very heavily on the origin of the string, as that would determine the special cases that needed to be dealt with. Can you provide more details?

edit: Based on what you said in your original post, I would say to have a list of possible delimiters (which would probably need to be added to for some time), and tokenize the string according to that, and discard any token that appears in a second list of words that don't matter (conjunctions, articles, prepositions, etc...). Before discarding said strings, you'd also want to check if they're operators used in your app, or anything like that.

gopher 16 years ago | |

I think of something like a comment or an abstract; security is not an issue here because input validation and escaping is done elsewhere.

Basically, I think of a string like "ham, egg." which should result in "ham" and "egg", and "Ветчина, яйцо." should also result in "Ветчина" and "яйцо".

The challenge is that you cannot whitelist all possible characters as there are (imho) too many charsets.

TallGuyShort 16 years ago | | |

Well barring the practice of specifying meaningful characters, the only thing I can come up with is to have your program use statistics to take it's best guess at what 'special' characters are. Let's say 95% of the characters are between 65 and 90, and every now and then there's a 44-32 pair. Then your program could be pretty sure that 44-32 is a delimiter, and that 65 and 90 are the ranges of characters used in keywords. (The above examples are ASCII).

However, that does nothing to eliminate words like 'in' and 'of' in a query, which you may want to do. It isn't very practical at all, I think, and you probably want to look at more practical ways to list possible delimiters, etc... Although the above could help you determine what charset you're using.

dannyr 16 years ago |

How about term extraction?

http://developer.yahoo.com/search/content/V1/termExtraction....

mbrubeck 16 years ago |

"Second trial, you use a alpha-numeric whitelist and split on anything else, but what about umlauts? What about hebrew or cyrillic?"

A multi-lingual version of this could use the Unicode "General Category" character classes (Letter, Mark, Number, Punctuation, Symbol, Separator, Other).

alanthonyc 16 years ago |

Not sure what your main goal is, but in my compilers project class, we used lexical analyzers to break out tokens from the input stream.

Try looking up "Lex" or "Flex"...these were the tools we used. There may be better ones around now.

Here's a quick google: http://dinosaur.compilertools.net/

pedalpete 16 years ago |

I just found LingPipe on Monday, and haven't had a chance to try it yet. But it has 'entity extraction' in text mining. Not sure if that is what you're looking for. It's a Java library. http://alias-i.com/lingpipe/

anybody have any comments about it?