Yeah I think what u need is one Filed where you store a list of propertytag and value combination and also be able to search on the filed on values and identify that the value is for a particular propertytag.
something like propertytag1, value propertytag2,value propertytag3,value etc To be frank I do not know how to do it. I suggest you post it as separate question so that people do not have to follow your thread. Here are some wild ideas look into customer analyzer stuff. You may have to write a customer analyzer to analyze such a filed. So, the idea is Construct a filed and give it a set of propertytag and value combinations analyze it so that you donot parse the pairs i.e each pair is intact and ready to search and retrieve it as a pair not as individual words. On Jan 14, 2008 10:33 AM, Roger Camargo <[EMAIL PROTECTED]> wrote: > Why am I afraid? > Building the index wouldn't be a problem. I guess. > Querying it would be more difficult. > > Let's see. > > Custom properties... defined by the user, there is no restriction, in > quantity, and values. > > > Custom property name: Frequency> > Custom property value: Quarterly> > >> > Custom property name: Last update> > Custom property value: 2006-01-01 > > Cube1 has Frequency, Last Update > Cube2 has Suggestion, Frequency > Cube3 has Time Range, Original Source > ... > ... > > In the index the document should have one field per CustomProperty > > docid=0, cubename=cube1, Frequency=Quarterly, Last update=2006-01-01 > docid=1, cubename=cube2, Suggestion="This cube should be used by the > Account manager", Frequency="Monthly" > docid=2, cubename=cube3, Time Range="Since 1980 up to 2007", Original > Source="Account subsystem" > .... > .... > > Each custom property should be a Field, there is no control over which > fields are created. > How the query should be constructed, at the query you specify which fields > you should use. There is a way to query all the fields. > > Well it looks like it should work. > > Is this how I should do it? > > Add each Custom Property as a Field is the answer? > > For the hierarchy info modeling, the answer is generating every > combination of dimension and levels? > > Thanks for taggin me along with this. > > Roger CamargoUMSS "University"Cochabamba - Bolivia > > > > Date: Mon, 14 Jan 2008 09:56:36 -0500> From: [EMAIL PROTECTED]> To: > java-user@lucene.apache.org> Subject: Re: How to model hierarchy info to > be searched related to a document> > I am not sure why you are afraid of > adding more fields to the document.> Having 20-30 fields to a document is > not a bad thing to do. Do you have any> constraints to limit the number of > fields in the document?> > > > > > > > > > On Jan 14, 2008 7:59 AM, Roger > Camargo <[EMAIL PROTECTED]> wrote:> > > Thanks for answering.> >> > > It seems that there isn't any other way around, having every combination> > > of dimension and level.> >> > The example for the observations of the > dimension, would be as follow,> > maybe isn't such an important information > to be stored, but type it is.> >> > Dimension name: Region> > Dimension > observations: The dimension only includes countries of south> > america > <--This would be the observations of the dimension.> > Dimension type: > Geographical> >> > OK, now it seems that I have one document per combination > of dimension and> > level (following the example I provided it would be > something like this> > doc 0: id=0, name="Quarter sales per region", > desc="Description of the> > cube", dimension="Region", level="Country"> > > doc 1: id=0, name="Quarter sales per region", desc="Description of the> > > cube", dimension="Region", level="City"> > doc 2: id=0, name="Quarter sales > per region", desc="Description of the> > cube", dimension="Time", > level="Year"> > doc 3: id=0, name="Quarter sales per region", > desc="Description of the> > cube", dimension="Time", level="Quarter"> >> > > OK and adding the Dimension type would end up like this:> >> > doc 0: id=0, > name="Quarter sales per region", desc="Description of the> > cube", > dimension="Region", dimensiontype="Geographycal" level="Country"> > doc 1: > id=0, name="Quarter sales per region", desc="Description of the> > cube", > dimension="Region", dimensiontype="Geographycal" level="City"> > doc 2: > id=0, name="Quarter sales per region", desc="Description of the> > cube", > dimension="Time", dimensiontype="Time" level="Year"> > doc 3: id=0, > name="Quarter sales per region", desc="Description of the> > cube", > dimension="Time", dimensiontype="Time" level="Quarter"> >> > Now what about > the custom properties?> >> > Custom property name: Update Frequency> > > Custom property value: Quarterly> > Custom property name: Last update> > > Custom property value: 2006-01-01> >> > I'm afraid that I would end up not > using them for the search, they would> > have each one a field, and because > the user can create any number of these> > custom properties they would be > hard to query them. Or is there any correct> > way to add them to the > document model?> >> > Roger Camargo> > UMSS "University"> > Cochabamba - > Bolivia> >> > ------------------------------> > Date: Sun, 13 Jan 2008 > 17:07:52 -0500> > From: [EMAIL PROTECTED]> > To: > java-user@lucene.apache.org> > Subject: Re: How to model hierarchy info to > be searched related to a> > document> > CC: [EMAIL PROTECTED]> >> > >> > Roger,> >> > Why can;t you have one document for every combination of > dimension,> > level ? Add cube name , id and description too as a field to > all documents> > , all it would be reduntant information, but you can live > with it i suppose?> >> >> > I think you are developing an application to > search a cube ?> >> > what do you mean by observations of a dimension ? is > there an example?> >> > On Jan 11, 2008 7:57 PM, Roger Camargo < > [EMAIL PROTECTED]> wrote:> >> > I'm trying to index information > related to Olap Cubes.> >> > Each cube I'm trying to model it like a > document.> >> > The cube have the following information:> >> > ID - Unique > identifier for the cube> > Name - Name of the cube> > Description - > Description of the cube> >> > (There can be many dimensions per cube)> > > Dimension name - Name of the dimension of the cube> > Dimension observations > - Observations related to the dimension. (Optional)> > Dimension type - Type > of the dimension> >> > (Each dimension has at least one Level, but there can > be many levels per> > dimension)> > Level name - Name of the level> > Level > observations - Observations of the level (Optional)> >> > (There has to be > at least one Fact per cube)> > Fact name - Name of the fact> > Fact > aggregation - Aggregation of the fact> >> > (Also there can be custom > properties added by the user, with the form> > Name,Value)> > Custom > property name - Name of the custom property> > Custom property value - Value > of the custom property> >> > Right now I'm just indexing the first 3 ID, > Name and Description, but I> > would also want the other information to be > indexed and search the cube with> > that information.> >> > > --------------------------------------------------------> > Data sample:> >> > > ID: 0> > Name: Quarter sales per region> > DescriptÃon: Description of the > cube...> >> > Dimension name: Region> > Dimension observations: The > dimension only includes countries of south> > america> > Dimension type: > Geographical> >> > Level name: Country> > Level observations: Observations > of the level....> >> > Level name: City> > Level observations: Observations > of the level....> >> > Dimension name: Time> > Dimension observations: Has > data only from the year 2000> > Dimension type: Time> >> > Level name: Year> > > Level observations: No observations> >> > Level name: Quarter> > Level > observations: No observations> >> > Fact name: Sales> > Fact Aggregation: > Sum> >> > Fact name: % Quarter Growth> > Fact Aggregation: AVG> >> > Custom > property name: Frequency> > Custom property value: Quarterly> >> > Custom > property name: Last update> > Custom property value: 2006-01-01> > > --------------------------------------------------------> >> > My problems > would be the following.> >> > 1. How to index "Dimension name" and > "Dimension observations".> > If there would be just Dimension names, I cound > index it as a single Field> > with multiple values.> > But with the addition > of the observations, I need to know if the search> > term was founded within > the observation, to wich dimension belongs the> > founded observations.> > > And the same happens with the "Dimension type"> >> > 2. There can be many of > these Dimension name, observations, type. The same> > applies for the Level > name, observation - Fact name,type - Custom property> > name, value.> >> > > 3. The levels. if the search term was founded in the level observation, I> > > would need to know to which level name is related the level observation> > > founded along with the dimension related, and finally the cube itserlf.> >> > > Well... this was a bit long question to be my first one.> >> > Maybe what > I want can't be done, maybe there could be some walkaround that> > someone > knows it.> >> > I was thinking that if it can be posible to have a field, > with additional> > info attached to the value, that is not searchable, it > just needed when the> > field value is retrieved it.> >> > For example a > Multi-value field called DimObs.> >> > Value1: "Observations related to the > first dimension of the cube"> > Related info: "Dimension name1"> >> > > Value2: "Observations related to the second dimension of the cube"> > > Related info: "Dimension name2"> >> > When the search is performed and is > founded in the DimObs, for example> > "first".> > Then the search found > "first" int the DimObs, but I would need to retrieve> > the "Related info" > to know to which Dimension belongs the observations> > founded.> >> > Thanks > in advance for keeping with me till the end of this mail and for> > any > suggestions that could give me.> >> > Roger Camargo> > UMSS "University"> > > Cochabamba - Bolivia> >> > > _________________________________________________________________> > Express > yourself instantly with MSN Messenger! Download today it's FREE!> > > http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/> >> >> >> > > ------------------------------> > Express yourself instantly with MSN > Messenger! MSN Messenger< > http://clk.atdmt.com/AVE/go/onm00200471ave/direct/01/>> > > _________________________________________________________________ > Express yourself instantly with MSN Messenger! Download today it's FREE! > http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/ >
