Skip to content
Aurélien Bénel edited this page Jun 27, 2017 · 14 revisions

What is the minimal set of Hypertopic tools needed for text analysis?

  • LaSuli is the "client" which will enhance your web browser so that you can annotate web pages with categories,
  • Argos is the "server" in which you will store your annotations to retrieve them later from other computers or to share them with colleagues.

Why is it usually not enough?

  • When you get more and more documents, you will soon need a synthetic view of your annotations. This client, acting as your portfolio, is Porphyry.
  • With a growing expertise in qualitative analysis and in your corpus, you will notice in the texts "markers" of a category. For example, in French, soi-disant, censé, aurait, pseudo could be defined as markers for the criticism category. Cassandre will allow you to define such "notional families" (cf. Luhn). For performance reasons, Cassandre requires that you store your texts in it, therefore it will be used also as your texts archive server.

Can I use both annotations and notional families on the same corpus?

Not for now, but we work very hard on this feature. The new protocol which will allow this is already in place for the communication between LaSuli/Porphyry and Argos.

Meanwhile, you can still use notional families with the previous versions of Porphyry and Cassandre.

How do "analysis frameworks" and "categories" in LaSuli relate with the Hypertopic model?

"Analysis frameworks" are Hypertopic "viewpoints", and "categories" are Hypertopic "topics".

Hypertopic (and the former Porphyry model) was designed long before we thought of using it for sociology and qualitative analysis. Moreover, it is usable in very diverse domains where "viewpoint" and "topic" are relatively easy to understand.

In LaSuli, which is more specific to qualitative analysis, we had to change the terms "viewpoint" and "topic" because they were misleading (especially in sociology).

What should I do if I experience bugs?

Don't panic, bugs are inherent to software. The good news is that there is a high probability that these bugs will be "known bugs". This means first that their fix is already scheduled, and secondly that we found workarounds for them.

In particular, you will avoid known bugs if you carefully follow these simple rules:

  1. While LaSuli is on, keep only one Firefox window opened.
  2. If your login or password was wrong, don't type them in the dialog. Go back to the settings panel to set them up again.
  3. As soon as LaSuli displays an offset in highlights, refresh the Web page.
  4. Close LaSuli, before you visit Web sites unrelated with your analysis.