We have added some features to our website so alarms can be classified by type.
Many of you are familiar with NFIRS codes. Codes starting with 100 mean “fire”. For example:
NFIRS 162. Outside equipment fire. Includes outside trash compactors, outside HVAC units, and irrigation pumps. Excludes special structures (110 series) and mobile construction equipment (130 series).
NFIRS is a US standard, but we have modified our system to support other systems as well. Currently NFIRS is the only classifier that is programmed; if you want another classifier, you will have to let us know what your system or country uses.
- We have added a column on the alerts tab that lists the alert type for easy reference
- You can select an alert type by clicking on the alert and scrolling to the bottom of the window.
- You can do a full text search by name or by code number. We have also reorganized the codes into an easy-to-browse list. When you have made a selection, push “save”.
- “See also” suggestions, if any, will be displayed below the description. NFIRS codes are complicated, we have tried to make it easy to choose the right one.
- We are adding a server component to learn from your selections and auto-select types for future alarms. In order to train our server, we are asking users to log into their web consoles and manually classify a couple dozen recent alarms. We can use this data to perform machine learning.
Some users have pointed out that a machine can not be accurate 100% of the time, especially since an alert classification often changes based on what is found on scene. A “651 smoke” alarm might turn into a “111 Structure Fire” or a “733 Smoke detector malfunction”. This is reasonable. To ensure clarity, human classified alarms will be shown in dark letters with the full 3-digit code on the Alerts tab. Computer classified codes will show the century digit only and will be grayed out. For example:
111 Fire (black, 3 digits) = Human classified
100 Fire (gray italics, 1 digit) = Computer Classified
Of course, if the computer gets it wrong you can always correct it manually.
The next step (once alarms are being classified right) is for us to update the pie graph to show the new classification system.