Quickstart , more?

[jlewis]

I downloaded Wandora and worked with it for a while.

Read the Quickstart and much of the documentation, but I need to know the basics. HOW do the links get formed so a graph can be made? I extracted in a couple PDFs and htmls but all that shows up in a graph is the title. I'm looking to generate a semantic map. I did the same with a Bing search and had the API running so it worked, but the graph just shows one entry for the search title.

Can we get at least one example that includes bringing in data from a general source like a PDF or HTML or Bing search to final graph?

Let's say I run a Bing search, it gives me 20 items shown in the left pane, now what?

[akivela]

Hi JLewis

PDF and HTML page extractors really are simple and do nothing but create a topic for each file. File content is attached to the topic as an occurrence but file content is not used to create any links (you mean associations, I assume) between file topics. If you wish to link topics representing files, you need to apply other extractors provided by the Wandora application.

I think a very similar question was asked in [1]. The solution is to refine occurrences carrying file content. You might also want to read [2] which shows a step by step example -- chapter Batch refining example -- how Calais is used to extract associations out of text occurrences.

About the Bing search question. Your next steps really depend on what you are trying to achieve. Again, you could try to refine occurrences in search result topics if that is something you want to do. In order to answer more specifically I should need to know what you want to do.

Speaking generally, I would not underestimate to importance to creating links i.e. associations manually. You can always create any associations. Documentation is [3]. WandoraTV has also a video clip about it [4]. In general, Wandora is not trying to give you a final graph but tools to build and edit your graph. Extractors represent just one way of building a graph semi-automatically. It usually requires several tools and manual editing to get your graph sufficient.

Kind Regards,
Aki / Wandora Team


[1] viewtopic.php?f=3&t=91
[2] http://www.wandora.org/wandora/wiki/ind ... ccurrences
[3] http://www.wandora.org/wandora/wiki/ind ... ssociation
[4] http://www.wandora.net/wandora/tv/

[jlewis]

Thank you for your generous help.

I tried stepping through each of those links you sent. Unfortunately there are some very basic issues that prevent it. For example the techniques make heavy use of the Topics > Occurences > Refine tool , but that tool does not appear in my (Windows 7, 64 bit) application.

I have imported several text files and used the Alchemy API to extract keywords and this works well, but without the Refine tool I can't do the same for data that is already in a Topic. My goal really is to mine documents and feeds to wrap my mind around what topics and interconnections they contain, so relying on intimate knowledge and manual creation of keywords and links would require me to accomplish all my goals before starting. The Alchemy keyword extractor, when applied at the point of import, gets close to the effect I'd like, but once I do something like import an RSS feed (great tool) , I then can't run Alchemy on it because the Alchemy tool is only in File > Extract!!!

Even just having a way to run the Alchemy tool on an existing topic would be very useful. Some of the examples seem to suggest this can be done with the Refine tool, but again that tool does not seem to appear in the version I downloaded and ran. Perhaps the tool only appears when certain qualifications are met, but I don't see that in the documentation.

==EDIT===
Okay, the issue was that I did not understand what was meant by "occurrence" in the documentation. Typically that is pretty deep and doesn't contain much text when I run RSS extractor or otherwise. Now I can refine if I can get text into the occurences.

Next challenge, creating topics! First try was to click "New Topic" and put in some information. Nothing shows up under topics after that though. I'm sure this software is great once one learns where everything hides!

Thank you again for your help.

JL

[jlewis]

Okay, so I've now created topics and pasted text from the PDFs into occurences and used Alchemy to keyword them. This is obviously going to be a whole lot more complicated than just pointing the tool at a directory full of PDFs

Is there any way to have the PDF or RSS extractor actually pull in the text and dump it in an occurence that one can refine? I can do everything I want manually by copying and pasting all the text into occurences and then keyword extracting, is there any way to automate that sort of thing or is it really manual?

EDIT--

For some clarity on what I'm trying to accomplish look at the "Semantic Live Map (DEMO)" On the lower left of this page: http://www.alchemyapi.com/

My goal is to be able to generate something like that from archives of documents.

[akivela]

Hi JLewis

I will try to answer your questions. Please notify if I somehow manage to miss a question.

First, finding the missing menu Topics > Occurences > Refine. I have a feeling that you have been looking for it from the top menu bar. It doesn't locate there. You'll find it in a popup menu opened in context of a topic selection. For example, right mouse click the Wandora class in the left column, in the beginning of topic tree. Locate submenu Occurrences. You'll find Refine there. I usually start refining in context of an instance table selection. [1] shows you a screen capture where the user is selecting to refine instances.

Second, creating and missing topics. If you create a topic with an option New topic..., Wandora doesn't connect you topic with any other topics. In other words, created topic has no associations, no classes (types) nor instances. It will be like a separated island. To locate such an island topic, use Finder tab or Search dialog. If you wish to make your topic connected in the very beginning, use topic creation options New instance topic... or New subclass topic.... Created connection depends on menu option and the context in which it was triggered.

Third, batch importing PDF files into Wandora. You can use File > Extract > Simple files > Simple document extractor... to import a complete folder containing PDF files into Wandora [2]. When finished, you can continue with refining occurrences of Document instances.

If you are using the AlchemyAPI to extract keywords out of PDF texts, notice the AlchemyAPI limits the size of posted text documents to 150 kilobytes. Thus, refining very long texts may generate errors.

I think the key issue is you to locate the Topics > Occurences > Refine menu. After that you can batch refine any occurrences you have otherwise managed to import into the Wandora. And repeating myself, please look at popup menus triggered with the right mouse button, in context of a topic selection.

Just one more note. Be aware of Wandora's automatic topic merge. Wandora merges topics automatically whenever they have same basename or subject locator or share at least one subject identifier. If you are not familiar with Topic Maps, the merge can cause you a trouble when it happens accidentally. On the other hand, it can be your friend too when you are incrementally building a topic map.

Kind Regards,
Aki / Wandora Team

[1] http://www.wandora.org/wandora/wiki/ind ... ple_04.gif
[2] http://www.wandora.org/wandora/wiki/ind ... _Extractor

[jlewis]

Thank you very much.

Your directions have resulted in me creating topics and refining occurences. Next project is that extraction you mention, which I will try today!

[akivela]

Great! Do not hesitate to ask a question if face problems in using Wandora.

Kind Regards,
Aki / Wandora Team

[IHSEN]

I dowloaded java and wandora but i can't start with wandora

[akivela]

Hello IHSEN

You message contains no information about how you can't start Wandora. To help you a little further, please tell, how you can't start with Wandora and what operating system are you using? Can you locate Wandora's startup scripts in bin folder? Does Wandora give you any messages or other visual clues, how it fails?

Kind Regards,
Aki / Wandora Team

[IHSEN]

thank you very much. now it is ok, i can start with wandora, but i have a collection of documents and i want to use wandora to generate the topic map of each document? if the document is text format for example how i do this

[akivela]

Ok.

To make a topic for each document you might want to try either Directory structure extractor or Simple document extractor. First one creates a topic map that blueprints your folder hierarchy. Latter creates a topic map where each given document has a topic. Topic contains file content as an occurrence. Latter is better, if you want to proceed in applying some other extractors to the file content. Notice, file content should be text. Both Directory structure extractor and Simple document extractor locate in Wandora's File > Extract > Simple files menu. You can give both extractors a directory and the extractor reads all files in the directory.

Also, some other extractors take files as input, and eventually create a topic for each input file. For example, GATE Annie extractor can read a collection of files and extract entities out of text files.

Please, check the documentation at http://www.wandora.org/wiki/Documentation#Extractors and http://www.wandora.org/wiki/Refining_occurrences too.

Kind Regards,
Aki / Wandora Team

[akivela]

Also, there is an old forum post that may be useful:

http://www.wandora.org/forum/viewtopic. ... 03dc8ef26c

--Aki

[IHSEN]

Thank you very much again for your help AKI