Extracting Keywords from DCRs

joyedele

I am currently trying to generate keywords for third party news articles that are pushed into TS via a web service. These articles are sent in via xml and transformed into the appropriate dct format so that editors can view the articles in TS. I've been successful in creating the index to generate the keywords for the articles, however it is generating it based on the entire dcr. Can someone provide any guidance on how to generate keywords based on one or two dcr fields (i.e, body and summary). Note: I need to ability to generate on these two fields b/c some articles do not come with a body; sometimes they just have a summary and an url that links to the actual article. Thank you in advance for your help.

Migrateduser

The result you are looking for is to have ONLY the content you want summariazed in the crackedText file as MetaTagger summarizes the articles. There are a numbrer of ways to achieve this. A simple way is to create a program in your preprocessor that captures just the fields you want to summarize. As the program ends, have it overwrite the content of the crackedText file with the values of the two fields. There are sample .pl scripts in the $mt-home/conf/processors/CLT folder. At the top you can see how the crackedText and originalText files are typically handled.