I need to come up with a design of a database handling research data from studies on emerging ventures in the Mobile Internet business, containing interviews with company representatives and financial data about companies.

Being a complete beginner on dB design and implementation, I turn to you guys here and hope that someone can give me some help or pointers to where to look for more information. I've browsed through material from university classes in dB and searched around on the internet, but I lack any "proper" skills.

I really just need the design, although I would be nice to be able to implement it (Oracle 10 and MS Access available) for testing purposes. The design will serve as information on our needs and is part of a larger system which IT-consultants eventually will design and implement, but I want give them an example on what we're aiming at (plus that it gives an opportunity to save some money on consultant fees...).

---

Our data:

A set of interviews with a number of companies is called a Survey, and a survey can contain:
* Interviews (BLOB) with [1..n] companies.
* Other material not connected with a specific company (BLOB) - for example information about the background of the survey.
* Tabular data regarding companies.
* Tabular data not related to a specific company.

Each company can have the following types of information:
* Interviews from [0..m] surveys (BLOB).
* Material not connected with any survey (BLOB) - for example a business plan.
* Tabular data relating to [0..m] surveys.
* Tabular data not related to a specific survey.

---

Based upon this I thought that I could have one table of BLOBs and a number of data tables - as there exists diverse tabular data and there is a future need of importing new tabular data from other databases.

I set out designing in DBDesigner (png, xml), using few tables where there would be entries referencing both Survey and Company - but there would also be a lot of entries only referring to only one Survey or Company. I came to wonder about the [0..*] relations. Is there any drawbacks using these?

Another possibility would be using more tables, dividing the information so that there are no [0..*] relations (png, xml).
Would this be more advisable?

Finally, I guess that there are several more ways to go about this. As stated earlier, any comments/suggestions are welcome.