ShowVoc confiugration and searching mechanism querry

[Md Monirul Islam Ansari]

Dear VocBench Community,

                                             Warm wishes to everyone and thank you for your kind support, through out this point.

Recently, we are trying to connect Vocbench and ShowVoc in one local machine(Linux/Ubuntu).
   We are using VocBench and ShowVoc in two separate machine, because of the problems in running same macine. (We are using Linux/Ubuntu OS, running $/semanrticTurkey/bin/start )
     

   We are able to install showvoc on another machine and configured vocbench which is installed on another machine

by providing these credentials. We first exported .ttl file from VocBench and created a project on showvoc and imported .ttl file.
   Is there any other method, we are missing ? How to connect vocbench and showvoc so that we can querry by realtime, without export and importing each time. ?

 

Another query, after we were able to run the showvoc and imported the project, When we are searching in "Global Search" , we needed to search by full term of a letter, Example: If we want to search 'femiboys' by entering word 'femi', it should pop up. It is working on admin interface, but in "Global Search" it is not working.

    Is there anything is missing which we are missing ? We are not so familiar with technical specialities,

if anyone found any way out ? please share with us.

Heartfelt thanks, and best regards,

Md Monirul Islam Ansari,
Student, Department of Library and Information Science,
University of Kalyani,
India

[Parthasarathi Mukhopadhyay]

Dear all

 

Let me add a bit here what Md. Monirul Islam Ansari is possible wish to communicate (as we were together in this project):

 

1. We failed to start ShowVoc on the same machine with VocBench (versions are compatible) even after changing the port to avoid port collisions;

 

2. Thereafter, we installed ShowVoc on a different machine (in the same LAN and IP range), and this time it worked like a charm.

 

3. The test connection with the backend VocBench (from his laptop) reported success;

 

4. Initially, we thought that it would fetch data from the backend VocBench, but we were not able to do that.

 

5. Then we exported the vocabulary from the backend VocBench (RDF serialization format, TTL) and imported it to ShowVoc, and after indexing, etc., it is working fine.

 

6. The only issue we faced was that it is supporting exact term search (e.g., gender) and truncated search (e.g., gen retrieves gender, generation, etc.) in the admin dashboard interface, but the same truncated search (e.g., gen) is not working in the global search interface, though it is working for exact term search (e.g., gender).

 

7. An important point for us is to support multilingual search, and this is working for both the admin search interface and the global search interface (exact search only). Our vocabulary is a translation of Homosaurus in Bengali and Hindi (two major languages spoken in India).

 

In view of the foregoing, we have two questions:

 

1. Is it possible to directly interact (searching, browsing, visualization, etc.) from the backend VocBench without exporting or importing as described in point 5?

2. Is there something that we are missing during indexing to support truncated (we attempted with the wildcard like *, $,? but no effect) searches in the global search interface?

 

Regards

Parthasarathi Mukhopadhyay

[Tiziano Lorenzetti]

Dear all,

regarding the possibility to have both VocBench and ShowVoc dealing with the same data, you have two possibilities:

1. rely on the same SemanticTurkey instance: as you know both VocBench and ShowVoc are released as .war deployed in SemanticTurkey. You could deploy both war into the same instance of ST, so basically they will share the same exact backend and thus the same data. Anyway, without digging into details, this solution is highly discouraged, so the following is recommended.
2. have the data on a remote triple store (e.g. GraphDB) and access the same repo from both VocBench and ShowVoc. For example, you can create the project from VocBench using the "CreateRemote" repository access (read project creation documentation here, or see the quick test drive here), and then create a dataset in ShowVoc using the option RepositoryAccess: AccessExistingRemote, pointing to the same triple store and repository ID used by the other project in VocBench.
   In this way you still have different ST instance but they will both use the same data (hosted on the remote triple store)

About the search functionality: the search in the Data page is based on a sparql query that is dynamically built and thus allows the search to be customized (e.g. the string match mode "startsWith"/"Contains"/"Exact"/..., or the search mode "Search in URI/local name/notes", and so on) and thus is more flexible. The Global Search, instead, currently is more strict and search only matching exact tokens (which are indexed labels).

Note that since the Global Search is based on indexed data, if you intend to use ShowVoc on the same data that is edited in VocBench, when you add/edit/delete labels in VocBench, you need to re-index the dataset from ShowVoc so that the changes are reflected into its index (otherwise the search could return outdated results).

Regards,

Tiziano

--
You received this message because you are subscribed to the Google Groups "vocbench-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to vocbench-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/vocbench-user/CAGM_5ubXtKpbKg9wvAQofA91dQPjScy2seGQD-zMAG70C3C6cA%40mail.gmail.com.

[Parthasarathi Mukhopadhyay]

Thanks Tiziano.

It is now working by using the backend GraphDB (Free) on the basis of the guidance that you provided.

We would like to explore translation of ShowVoc in Bengali and Hindi but the instruction given here - https://showvoc.uniroma2.it/doc/sys/internationalization.jsf instructs

"The translation file needs to be placed under assets/l10n/ folder where those already existing are. Finally it is necessary to "inform" ShowVoc that a new language is available so that it will be then listed in the language selector. This can be done by editing the svconfig.js file (under src/ of the source package, or under the root folder of the built distribution) "

but in our extracted folder (showvoc-2.4.2-full) we are not getting any such directory (asset) or file (svconfig.js).

Regards

[Tiziano Lorenzetti]

Dear Parthasarathi,

you're right, the documentation is a bit ambiguous on that point. For "built distribution" we mean the .war of ShowVoc (located under semanticturkey-<vers>/deploy/) , so you can:

1. Extract the content of the .war archive
2. Edit svconfig.js by adding the desired languages to the variable additional_l10n_langs 
3. Add the translation files under assets/l10n/ (I suggest you to start by editing one of the existing translation files, e.g. en.json)
4. Repackage all as WAR archive and deploy it in place of the existing showvoc-<vers>.war

Best regards,

Tiziano

[Parthasarathi Mukhopadhyay]

Dear Tiziano

Thanks again.

We got it.

When you say : Repackage all as WAR archive and deploy it in place of the existing showvoc-<vers>.war, it reminds that we did something similar when integrated VocBench generated vocabulary with Loddy (based on this FAQ - https://vocbench.uniroma2.it/documentation/faq.jsf).

We need to explore and report.

Thanks and heartfelt regards

Parthasarathi

To view this discussion on the web visit https://groups.google.com/d/msgid/vocbench-user/CAOA09s1Pr%3DApJnPNZE2O5W8CVF9wCFKoE-bdJZVLaTv8Qa1gDA%40mail.gmail.com.