feixiangcq wrote: > Hello, > I have homework about designing AE in UIMA AS.The AE includes there steps: > 1.It reads data from file system; > 2.Build a model from the data just readed; > 3.Analyze the CAS which I get from CollectionReader using the model in step 2. > If I deploy the AE on a computer and analyze one CAS,it works well. > However,If there are many CASes and the AE can only analyze one CAS at a > time,what about the cost I spend? > That is,every time the AE anlyze a CAS,it must do step 1 and 2, > but actually the model in step 2 are the same for all CASes. > Step 1 and 2 are repeated and they cost much of the time. > Can the AE only need do step 1 and 2 once and it can analyze all CASes? > Or is there some sollution to reduce the cost of step 1 and 2?
You could do 1. and 2. at initialization time, by implementing the initialize() method. See http://incubator.apache.org/uima/downloads/releaseDocs/2.2.2-incubating/docs/api/org/apache/uima/analysis_component/AnalysisComponent.html#initialize(org.apache.uima.UimaContext) --Thilo > By the way,I have another question.Can the CollectionReader of computer A > read files from computer B? Is there an example in UIMA AS source code? Or > can > anyone give me an example? > I am very glad to hear you answer. > Thank you!
