|
SAT Home | About Us | Contact | CassBeth Products |
|
|
1. What should my specification format be prior to uploading
to SAT?
2. Should I have a PUI for my requirements?
3. Should I have requirement attributes?
4. How can I save my unique rules?
5. Is there a size limitation on my Specification file?
6. Are there sample specifications I can use for
download?
7. I like SAT but it is too expensive for me?
8. Where did SAT come from?
9. What is meant by "level "and "allow Humans to make final
decisions at each level"?
10. Please explain what is intended by "layering", "levels",
"services" and their relationships?
11. I get the impression that SAT has access to other
specifications, that can be consulted, when needed?
12. SAT will evaluate text on presence or absence of text using
rules, but signalling the defect does not create "good requirements", please
explain?
13. I can't play games with tool support, how can you meet
my needs for life cycle support?
14. Why is SAT so special?
15. Is there a reason for the colors in the SAT reports: Red,
Green, Blue, Purple means?
16. How do I install and administer SAT?
17. Why use SAT, I can do this using my eyes and some
quick searches?
1. What should my specification format be prior to uploading to SAT?
If your spec is a document, save it in text format. If your spec is in DOORS export it as an Excell Spread Sheet using .csv (comma delimited) format. Be careful to set your DOORS ID field to something else like PUI prior to the export otherwise you will be unable to open it in Excel. Apparently Excel uses ID in .csv format.
2. Should I have a PUI for my requirements?
Project Unique Identifiers (PUI) can be part of your file. If so remember to add the PUI mask in the SAT rules. The mask is currently set to REQ-\d+. If you have no PUI, SAT will add its own PUI with the preface of "REQ-".
3. Should I have requirement attributes?
Sure. Especially an attribute that identifies if an object is or is not a requirement. The SAT pre-defined rules use NotReq for the requirement attribute field. Attributes can be very valuable when creating your own unique rules.
4. How can I save my unique rules?
Just save your web page on your hard drive. When you want to duplicate your analysis, just load your saved web page into your browser and press the submit button prior to submitting a specification for analysis. You are now back to where you can duplicate your analysis.
5. Is there a size limitation on my Specification file?
The curent version is set to 20 Mbytes. Your solution can be set to 2^17, but something like 1 Mbyte is a practical size. Emperical data suggests that 1 Mbyte is 5000+ objects (requirements with several attributes in .txt format).
6. Are there sample specifications I can use for download?
Yes. They are part of the application and used in the help area.
7. I like SAT but it is too expensive for me?
If you are a one or two person shop out there consulting to feed your family, give us a call and we will work something out. This does not apply to small businesses, 8A companies, or corprations. This only applies to the lone group of people who may have s-corp status.
8. Where did SAT come from?
The answer is complex. One driver was a perceived need identified at a technical meeting in 2005. Our take way from the meeting were these thoughts - About SAT. Another driver was the technology developed by CassBeth and its Internet studies.
9. What is meant by "level "and "allow Humans to make final decisions at each level"?
Reviewing the spec is an iterative process and starts by selecting the most simple service as the first step. The second step is to tune the rules for this service if this is the first run in the organization or project (on the fly rule authoring). The third step is to select the actual offending objects and filter those not needing attention (check box filters). The third step is a pick your battles decision on the part of the analyst. Between each run the user is able to save the settings and results. These saved runs can be re-executed with new settings based on a new "day" or the results of a formal review.
The tool groups the rules into services. Each service has a specific goal. If all the services were enabled, the user would be overloaded with data. The idea of a "level" is based on ability to separately select each service and each rule within a service. There is also the concept of simple services and complex services implying a "hill" that the analyst climbs.
Finally, within each service the user is able to selectively pick each rule on the fly. So for the key req's service a safety engineer can disable all the other "rules" and just focus on the "safety key req's". For the req text analysis service the most offending and accepted rules can be selected first until the team understands the benefits, they learn, and the rule is rarely triggered, then the more "controversial rules" can be selected to make the req's even better. Again, this is the concept of levels.
All of the above is done on the fly in real time...
10. Please explain what is intended by "layering", "levels", "services" and their relationships?
Layering is related to services. Services range from conceptually simple (bad words) to complex (document shape). Services are made of one or more rules, and services are grouped into templates, and templates are further grouped into industries or subparts within each industry or project.
You can visualize each of these elements as a dimension in a cube.
So you have a cube that represents your project, then you have multiple cubes that represent multiple projects in your enterprise.
11. I get the impression that SAT has access to other specifications, that can be consulted, when needed?
The premise of SAT is to accept that different organizations have different rules and they are successful. The goal is to offer a tool that they can easily tune to their unique needs. This is accomplished with a default template that is easily modified by the user. This default template includes rule attributes that provide guidance when needed. This guidance can be stand alone or links to other objects using HTML code.
For example, if you have your own version of the PlainLanguage guidance document, you can place that link in a rule attribute when guidance is offered to the user.
12. SAT will evaluate text on presence or absence of text using rules, but signalling the defect does not create "good requirements", please explain?
Remember the premise of SAT is to accept that different organizations have different rules and they are successful. The goal is to offer a tool that they can easily tune to their unique needs.
Each rule has attributes. One attribute is a free text attribute, which is used to encode the knowledge of how to deal with the particular finding. If you know enough to create a rule you should know enough to provide guidance. SAT includes limited templates. The goal is to provide the user with enough information without invalidating the tool because of their unique needs.
For example, some organizations have moved away from "shall" to "must". The templates do not say you shall use "shall" because we think it is correct. Instead the templates provide a strong example, teach, and allow "must" as a primary imperative. SAT includes a help area to teach a user. There is also an audio presentation, which are used for initial training, when requested.
13. I can't play games with tool support, how can you meet my needs for life cycle support?
We realize our primary customers have long LCS needs that may range in the decades. We offer several purchase options including perpetual licenses and source code.
14. Why is SAT so special?
SAT operates in real time on-the-fly as you ask your questions. The rules are closely bound to the analysis results encouraging you to modify your rules when it makes sense. Its interface is based on the display filter concepts pioneered by air defense and air traffic control workstations allowing you to get a big picture view or extremely detailed view within seconds. Its operation and output is based on the Internet search engine concept making it very easy to understand, even though it performs amazing analysis.
15. Is there a reason for the colors in the SAT reports: Red, Green, Blue, Purple means?
Colors - No reason. It is all according to the user and how they want to group the results. Sometimes multiple rules in consecutive order within a service are really part of the same general topic and using the same color helps to group those topics.
16. How do I install and administer SAT?
SAT uses a standard Windows install wizard. The user is requested to read and accept the License. The install wizard installs SAT in a fixed directory. After installation 2 activation keys are entered the first time SAT is executed. This is a unique configuration for SAT operation. For ease of upgrade, SAT is downloadable off the Internet.
Currently there are few SAT administration needs. Once you complete the install, your computer department can copy the image and push the application out to your enterprise. If you decide to not use this approach the install is very fast and takes less than a few minutes, including the entry of the activation keys.
The user is alerted 14 days prior to the expiration of the user keys. During this time a link is offered in the application to update the user keys. If the user keys should expire, you will need to reinstall SAT. The installation will "refresh" the baseline files and you will be prompted for the activation keys the first time the application is started. You should not lose any data, but if you changed your default template it will revert to the SAT baseline.
If you wish to update the keys prior to expiration of the 14-day window rename the "pck.txt" file in the "sat" directory and proceed like this is the first time you are executing SAT. You can then push the "new pct.txt" file out to your network after successful activation of SAT on your staging area computer.
17. Why use SAT, I can do this using my eyes and some quick searches?
You could, but here is the problem, it will take you a very long time and your results will not be consistent or reproducible. For example on first examination you feel you may be able to perform the SAT Requirement Text Service manually, but it has 15 rules and 71 words or phrase patterns some of which search while others reject. This service takes less than 60 seconds on a 148-page document that contains 3900 objects. Manually duplicating the other SAT services like Document Shape is futile and that path would be quickly abandoned.
At a recent NASA conference on this subject Carnegie Mellon mentioned that they found that using an automated tool surfaced requirement issues significantly faster and was more thorough than using a typical review with humans. The humans are good at finding the domain problems in a specification but poor at finding the other problems that these kinds of tools surface.
CassBeth learned this in 1997 when we started this work. In fact we all know this, we just tend to forget that computers count and search very fast and never make a mistake but humans can create. We also tend to forget that consistently counting and searching still can solve many problems.
So using a tool like SAT will significantly clean up your specifications and free the staff to work on the difficult problems. The problems that only they can address with their domain knowledge and creative abilities.
The following is a potential time line for a SAT run:
| Time | Activity | |
| 90 seconds | Save requirements document in text format (.txt) or DOORS Spreadsheet export (.csv) 2300 objects with 3 attributes and 2 links Note: exporting to Word is significantly slower, Spreadsheet is fast |
|
| 05 seconds | start SAT | |
| 90 seconds | load DOORS export or text document into SAT and run Requirement Text Analysis | |
| 15 minutes | enable disable rules, modify create rules, accept reject objects Note: most people stop searching the Internet for a subject after 2 days the same will happen with SAT, it is not an endless loop |
|
| 15 minutes | run remaining SAT services | |
| 4 hours | coordinate with closest peers what did you get, I got this, well I got this, so what do we do |
|
| 2 days | buy in from all the stake holders if there is still disagreement after 2 days the problem is beyond SAT |
Purchase . Download . Want to know more