7

I've recently become interested in geo-extraction; I'm eventually going to work on a project where I'll want to extract street addresses from various text sources (almost exclusively segments of web pages).

That said, would questions about algorithms and approaches to geo extraction be on topic? They would be along the lines of say, trying to find patterns to identify address segments that would later be geocoded to get a latitude/longitude pair.

Note: I am not asking about how to get structured information from what is already known to be an address, but the step before that, finding an address in a body of text.

For example, see geoextractor:

http://geo.bookmark8.com/ie.py?url=http%3A%2F%2Fecthrowdown.com%2Ftournament-information%2F

||||||
9

The processes of recognizing addresses in text and (especially) cleaning and standardizing those addresses are a frequent part of GIS workflows. When a geo-extraction question is formulated to appeal to the "domain knowledge" of the community, then--regardless of what software or programming techniques might be needed for its solution--I would welcome it here on GIS and would hope that it would get a warm reception as well as good answers.

||||||
  • That's what I was looking for. Yes, I'm not going to have programming issues per SE, but I can't code anything if I don't understand the processes and algorithms, and I think that's where you guys can help. Much appreciated. – casperOne May 2 '13 at 13:52
6

Could go either way but I think StackOverflow might be a better place for this type of question, especially when you start talking about text extraction/correlation algorithms rather than ready-to-use APIs and services. Here are some examples from there:

There are however examples of similar questions on this site as well:

One site I came across that you might find useful which has examples of extracting and then parsing addresses from large amounts of text is http://www.address-parser.com/

||||||
  • 3
    Note: I am a moderator on Stack Overflow, so while this might be on topic from an algorithm point of view, it's not the programming that's the important part, it's accessing those with the domain knowledge to lead to a better solution. Don't get me wrong, I love programmers, but we aren't always domain experts; I'd rather go to the domain experts and benefit from their experience and then translate that into my domain (programming, where I am confident in my experience). – casperOne May 2 '13 at 11:44
  • The last example you gave is not constructive on Stack Overflow, so that's not a good example, as I'm not looking for a tool. My questions would be more along the lines of "I'm trying to extract Australian addresses from text, I'm using this mechanism, but my problem is <so and so>". This is not really a programming question per-se. – casperOne May 2 '13 at 11:46
  • That said, I'm conflicted, is it or is it not on topic? I honestly don't think that SO is the better place for questions of this nature, but if it's explicitly off-topic here, then I won't post it here, but if it is, I'd like to. Your answer doesn't indicate strongly either way. – casperOne May 2 '13 at 11:47
  • Finally, all of your examples are about parsing into structured pieces what is already known to be an address. Geo extraction is about the process of getting the address in the first place. See this for an example: geo.bookmark8.com/… – casperOne May 2 '13 at 12:19

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .