Visualizing time-dense episodes

1. A method comprising:

identifying, by a system having a processor, attributes in data records;

detecting, by the system, episodes corresponding to the respective attributes, wherein each of the episodes includes a respective collection of events represented by the data records that relate to a corresponding attribute;

scoring, by the system, the episodes, where scoring of a particular one of the episodes is based on relative time densities between successive data records of the particular episode;

producing, by the system, a visualization of at least some of the episodes to provide an alert of time-dense episodes, the visualization including graphical portions representing the respective at least some of the episodes, each of the graphical portions including graphical elements representing respective events in a corresponding one of the episodes; and

ordering, by the system, the graphical portions in the visualization according to scores calculated by the scoring of the episodes, the scores including a score calculated by the scoring of the particular episode.

2. The method of claim 1 , wherein detecting the episodes comprises performing real-time detection of the episodes.

3. The method of claim 1 , wherein the scoring of the particular episode is further based on an aggregate time distance among the data records of the particular episode, where the aggregate time distance is used to normalize the relative time densities.

4. The method of claim 3 , further comprising:

adding a given data record containing a respective attribute to the particular episode in response to the given data record having a time distance from a previous data record containing the respective attribute by less than the aggregate time distance.

5. The method of claim 1 , wherein the scoring of the particular episode is further based on negativity of sentiment expressed in the data records of the particular episode.

6. The method of claim 1 , wherein the scoring of the particular episode is further based on content coherence among the data records of the particular episode, where the content coherence is based on strength of association of terms with the data records of the particular episode.

7. The method of claim 1 , wherein the graphical portions include respective time-density tracks to depict relative time densities between data records of corresponding ones of the episodes.

8. The method of claim 7 , wherein the graphical portions further include sequences of the graphical elements that represent respective events of corresponding ones of the episodes, where each of the graphical elements has a visual indicator assigned based on a measure assigned to an attribute in the corresponding data record.

9. The method of claim 8 , wherein the visual indicator for a respective one of the graphical elements is selected from among a plurality of colors that correspond to different values of the measure.

10. The method of claim 1 , wherein the visualization further includes links between pairs of the graphical portions, at least one of the links indicating co-occurrence similarity, the co-occurrence similarity based on co-occurrence of attributes in the data records.

11. An article comprising at least one non-transitory machine-readable storage medium storing instructions that upon execution cause a system to:

identify attributes in incoming data records;

add data records to episodes that correspond to respective ones of the attributes, where adding a given one of the data records to a particular one of the episodes is in response to determining that the given data record has a time distance to a previous data record of less than an aggregate time distance among data records of the particular episode;

score the episodes, where the particular episode is scored based on relative time densities between successive data records of the particular episode and on the aggregate time distance; and

produce a visualization of at least some of the episodes based on the scoring to provide an alert of time-dense episodes.

12. The article of claim 11 , wherein the instructions upon execution cause the system to further order graphical portions of the visualization representing respective ones of the at least some episodes according to scores produced by the scoring, wherein each of the graphical portions includes graphical elements representing data records of a corresponding one of the episodes, and each of the scores is based on relative time densities between successive data records of a corresponding one of the episodes.

13. The article of claim 12 , wherein each of the graphical portions includes a label of the corresponding one of attributes, and a time-density graph to depict relative time densities between data records of the corresponding episode.

14. The article of claim 12 , wherein the scores are further based on negativity of sentiment expressed in the data records of the respective episodes.

15. The article of claim 12 , wherein the scores are further based on content coherence among the data records of the respective episodes, where the content coherence is based on strength of association of terms with the data records of each respective episode.

16. The article of claim 11 , wherein the instructions upon execution cause the system to:

add a link between a graphical portion for a first of the at least some episodes and a graphical portion for a second of the at least some episodes, wherein the link indicates relative similarity between the first and second episodes.

17. The article of claim 16 , wherein the relative similarity includes co-occurrence similarity that indicates that multiple attributes describe a common issue as the multiple attributes co-occur in the same data records.

18. The article of claim 16 , wherein the relative similarity includes content-based similarity that indicates that multiple attributes are used synonymously to describe a common issue.

19. A system comprising:

at least one processor to:

identify attributes in data records;

detect episodes corresponding to the respective attributes, wherein each of the episodes includes a respective collection of events represented by the data records that relate to a corresponding attribute;

score the episodes, where scoring of a particular one of the episodes is based on relative time densities between successive data records of the particular episode;

produce a visualization of at least some of the episodes to provide an alert of time-dense episodes, the visualization including graphical portions representing the respective at least some of the episodes, each of the graphical portions including graphical elements representing respective events in a corresponding one of the episodes; and

ordering, by the system, the graphical portions in the visualization according to scores calculated by the scoring of the episodes, the scores including a score calculated by the scoring of the particular episode.

20. The system of claim 19 , wherein the at least one processor is to further add a given data record containing a respective attribute to the particular episode in response to the given data record having a time distance from a previous data record containing the respective attribute by less than the aggregate time distance.