Thanks Gora for a quick response. The catalogue views are not defined in the structure of the XMLs. The input to the Solr engine will be a base catalogue system comprising of the entire set of products. It is totally left for the search module to meet the business requirement.
For your options mentioned: 1. Different Solr Core would not be possible as these views are not defined ahead of time in XML. 2. FOr the second option to have it in multi valued fiels - WIll it be appropriate to have this field include incredibly large number of values for the record ? Say 1500 values for that field..? On Sat, Oct 6, 2012 at 9:04 PM, Gora Mohanty <g...@mimirtech.com> wrote: > On 6 October 2012 20:31, nitin malhotra <nitin.rs...@gmail.com> wrote: > > > > Hi! > > > > I have a business case to Index the Product Catalogue data using Solr > for a > > food chain client. The data received is in the form of XML. And the > > business requirement for search is that a log in customer on the portal > > when performs a search, it should be able to search only a specific set > of > > products from the catalogue. This can be understood as forming different > > views of product catalogue which have certain products available for > search > > and purchase. Also the products in a particular view (catalogue group) > can > > exist in more than one set. > > > How are the different catalogue views defined? Is this definition > included in the structure of the XML? > > > > > The # of views (or groups to be formed) - 1500 approx > > > > 1. Need help to understand the strategy to index data to be able to make > it > > available for search in the catalogue views. > > > Different approaches are possible depending on how your data > are structured: > 1. If the catalogue views are all defined ahead of time in the XML, > and if wastage of disk space from duplication of data in different > catalogue views is not an issue, a clean approach would be to > allocate a different Solr core for each catalogue view: > http://wiki.apache.org/solr/CoreAdmin > 2. Another approach would be to flatten all the data, and for each > product record have a multi-valued field that defines which catalogue > views the product should appear in. > > > 2. Is there a way to index the complete catalogue and then have some way > to > > search based on the product views ? > > > > Please go easy as I am new to the SOlr. > > The most important part is to design your Solr/Lucene index so as to: > (a) Make search efficient > (b) Make indexing easy in terms of the code that has to be written, > and make it efficient. > > If you are new to Solr, I would start by understanding the basics. The > Solr Wiki at http://wiki.apache.org/solr/ has a lot of information. > (a) Get Solr up and running > (b) Go through the basic examples, especially how to index XML: > - http://wiki.apache.org/solr/UpdateXmlMessages > - The Solr DataImportHandler can be useful here: > http://wiki.apache.org/solr/DataImportHandler > > http://wiki.apache.org/solr/DataImportHandler#Usage_with_XML.2BAC8-HTTP_Datasource > (c) After you are familiar with the basics, you can ask more specific > questions about how best to structure your index, providing > examples of the XML to be indexed and searched. > > Regards, > Gora > -- Have a nice day. Thanks and Regards, Nitin Malhotra
