Digital Mirror uses linguistic analysis of your email to identify topics. It then assigns a score to each topic to find the best candidates for display. The score is based on the degree of attention you appear to pay to the topic, as measured using parameters such as

• How frequently you reply to messages about the topic
• The typical length of your emails about the topic (longer is better)
• Whether you tend to store messages about the topic
• Whether the topic is associated with a "good" group of correspondents.

Digital Mirror selects a combination of groups and topics that displays a high level of activity over time.

Digital Mirror looks for groups of names that frequently occur together in the headers of your emails. It then determines the best candidates for display based on:

• How distinctive the group is - i.e. groups few of whose members are members of other groups are preferred
• How frequently you communicate with the group
• Over how long a period you have communicated with the group
• Whether the group is associated with a "good" topic.

Digital Mirror selects a combination of groups and topics that displays a high level of activity over time.

Digital Mirror creates groups that include everyone who participated in a conversation about a topic. But it is possible that not everyone is included in every single email about the topic. If that is the case, and Digital Mirror displays such an email in the evidence panel, that email may not involve every actor in the group.

Parts of the timeline at the top of the visualization may be colored yellow. This means that no data from the indicated period was used in this visualization. This version of Digital Mirror shows only a limited amount of information, so it is quite possible relevant data was found for this time, but is not being shown. (The timeline animation runs at higher speed through the yellow-colored regions.)

For this visualization, Digital Mirror looks for topics and groups of people that are associated with the greatest amount of discussion. The topic / group pairings with the most discussion are selected for display. Frequent recurrence over time is a factor in selection: a topic that was discussed frequently over a long time is more likely to be selected than one that did not last as long.

In this version of Digital Mirror, we select no more than five groups and five topics for display. If someone does not appear, it is probably because he or she did not participate in discussion of topics that extended over a long period of time, compared with others that were selected.

If you are running Digital Mirror on personal data, please note that this version is optimized for analysis of data in a work or other organizational context. For personal data, it is possible that you may find some of the results less interesting or relevant.

Even if that is not the case, people are sometimes puzzled by the results Digital Mirror presents. But these results are firmly based on the data: if you see something unexpected, it is likely to be because your data suggests something different than your perception. We strongly recommend that you look into the data yourself. In many cases, you can do this by clicking on the specific result in the visualization to bring up an evidence panel that shows some examples that led to the conclusion. In our experience, in the vast majority of cases, the data will support the Digital Mirror result. The challenge is to see yourself as Digital Mirror and others see you. If you have tried and still believe the result is wrong, please let us know.

Digital Mirror is designed to create a reasonably wide range of visualizations, not all of which will be applicable to everyone. It is quite likely that, for any individual user, there will not be enough relevant data for every single visualization. This is normal, which is why we provide a sample for you to enjoy!