Hi, My question is not strictly to do with image search but I can't help feeling the issue is somewhat related in terms of where to store what: I want to spell correct web pages prior to indexing and only index the corrected terms. I still want to store the original errorful text so this can be given in the summary and highlighted accordingly. Queries would similarly be corrected at query time. (The need for an alignment is to recover spelling mistakes and miskeys like "inthe" and "Nutc h" etc. although actually it's more relevant to CJK languages.)
Of course, this would be trivial if we didn't want to highlight the terms in the summary but to do highlighting we need to know where the original terms are in relation to the corrected terms (i.e what we indexed). We therefore need somewhere to store an alignment of the errorful original with the corrected text. At the moment we're storing this as an extra field inside the index but this is not a very elegant solution. Hence my related question: is there a more appropriate location to store such data? And where is the best place to do the spelling correction in the Nutch workflow? A separate mapreduce job or inside the indexing reduce job as I'm doing now? -Ed On 3/30/07, Doug Cutting <[EMAIL PROTECTED]> wrote:
Steve Severance wrote: > I am not looking to really make an image retrieval engine. During indexing referencing docs will be analyzed and text content will be associated with the image. Currently I want to keep this in a separate index. So despite the fact that images will be returned the search will be against text data. So do you just want to be able to reference the cached images? In that case, I think the images should stay in the content directory and be accessed like cached pages. The parse should just contain enough metadata to index so that the images can be located in the cache. I don't see a reason to keep this in a separate index, but perhaps a separate field instead? Then when displaying hits you can look up associated images and display them too. Does that work? Steve Severance wrote: > I like Mathijs's suggestion about using a DB for holding thumbnails. I just want access to be in constant time since I am going to probably need to grab at least 10 and maybe 50 for each query. That can be kept in the plugin as an option or something like that. Does that have any ramifications for being run on Hadoop? I'm not sure how a database solves scalability issues. It seems to me that thumbnails should be handled similarly to summaries. They should be retrieved in parallel from segment data in a separate pass once the final set of hits to be displayed has been determined. Thumbnails could be placed in a directory per segment as a separate mapreduce pass. I don't see this as a parser issue, although perhaps it could be piggybacked on that mapreduce pass, which also processes content. Doug
