3. Data mining area: Once mining is performed, it allows working with the obtained inference by applying different types of visualization or filtering.

4. Workflow graphical representation area: for visualizing the inference or inferences obtained as a result of applying the data mining algorithm.

5. Information area: Shows extra information related to the tasks status applied in the different sections.

1.2. Top menu

It allows the access to the application common functions. In the previous image, from left to right there are the following features:

1. Load button: allows searching the file containing the corpus data. (See section 1.2.1)

This button is not accessible in the “demo” version; the "Descargar Datos" button (“Download Data” in English) should be used instead (see right image).

2. Start data mining button: Starts the Data Mining Analysis. Previously, a data corpus and the desired filtering options must have been selected.

3. Save button: allows saving the mining inferred workflow and currently being displayed in a file (if you are visualizing more than one, it save the selected one). It is saved in the extension ".pmc" format. This saving allows its use in the function "Compare inference with a saved pattern" that is presented later.

4. Zoom button: to enlarge the inferred workflow drawn or, if there are several represented, the selected one.

1.2.1. Uploading your own corpus

When pressing the “Load” button, an “open” window appears. The user selects the .csv file to be loaded and click on the “open” button.

ATTENTION!! The .csv file must be closed in order to load it. If it is open, PALIA does not warn of any type of error. I just do not load it.

Now the “Asistente para la carga de datos” (Data Load Wizard) will appear, which will help to specify the loading parameters.

Now the type of parameters of the csv file fields that have been loaded must be specified. It could be done directly in plain text in the white box or by loading a .mcs extension file in which have been previously saved the configuration. The fields must be separated by the ";" symbol and can only be one of the following types: ID, START, END, ACTIVITY, RESULT, SAMPLE or EXCLUDE. The minimum required fields are an ID (at least one), an ACTIVITY (at least one), a START (only one) and an END (only one).

After pressing “Siguiente” (Next) a window appears to determine how we want to calculate the occupation:

At the moment we do not want to calculate occupation, so we will select the option “Omitir datos de ocupación” (Skip occupancy data).

When clicking “Next”, different options to save preselections are presented in the window.

If we are going to load the same .csv file many times and we foresee the options we have just selected will always be the same, we could use the first option “Salvar todos los datos combinados en un fichero único” (Save all the combined data in a single file). By clicking, it will ask to give a name to this file. This allows to have all the options saved in a single file, so to upload the work later we simply should click on the “Load” button and select the .smo file. This will load both the .csv and the options that we have selected for its load

On the other hand, if we foresee that we will only keep the structure or occupation data, it can also only be saved that option. When we finish, press "Next" and our .csv will be loaded in PALIA.

1.3. Filter area

The filters helps to adapt the input corpus data to the mining module format. In this area can be found the different filters to modify and select which initial corpus data will be processed in the mining engine. According to the chosen corpus, only the applicable filters will be loaded.

1.3.1. Filters: Top menu

In the upper part of the filter area there is a menu that allows to perform functions on the loaded data or on the selected filter options.

1. The first button allows to save the data of the loaded corpus in the ".smo" joint file format where are included the data of the selected corpus, the data of the file information structure and the occupation statistics data.

2. The second button allows opening ".flt" a file which contains filtering options already configured and loads them in the filters.

3. The last button allows to save in a ".flt" file the currently selected filtering options.

1.3.2. Filter: Levels

The different locations detected by the Sphera system are grouped into different levels. There are levels that contain the smallest detectable areas and levels that group different areas generating global location zones. For example, on a single level can be found the "Operating Room 1", "Operating Room 2", etc. areas and at another level, these areas are grouped within the "Operating Rooms" area.

This filter will allow defining with which granularity of area within the existing categories it is desired to work. Each of these areas will be represented with a node of the workflow. If more than one level is selected, the names of each level are concatenated to generate the nodes. Following the previous example, we would obtain "Operating Rooms_Operating Room1", "Operating Rooms_Operating Room2", etc.

1.3.3. Filter: Name Nodes

This filter presents the names of the nodes or location areas of the system. These names are dependent, as has already been explained, of the selected ones in the Level Filter.

This filter allows indicating which areas or nodes are desired to include in the mining, removing the checkbox mark of the areas desired to be excluded from the data corpus. This filter also allows renaming the area of location, being able to assign a new name to an area or even group two areas giving them the same name.

1.3.4. Filter: Dates

It offers the possibility of limiting the corpus data of a range of specific dates, so if it is a whole year corpus, the data of a specific month or week can be analysed.

1.3.5. Filter: time periods

With this filter, the activities/locations that were active in a specific time period are selected from the corpus, that is, only the location data included or intersecting in that time period are selected.

This filter separate the samples by days, that is, it takes the samples lasting several days, cuts them on each days and uses only those activities that meet the time period. Therefore, if this filter is applied to a several days corpus, several corpus will be obtained, where the same sample can be in several corpus and in each corpus will be the activities that met the indicated time period.

For applying this filter, the "Activar" box must be checked.

1.3.6. Filter: Duration

Allows selectin the activities/locations that last a specific minimum and maximum duration.

For applying this filter, the "Activar" box must be checked.

1.3.7. Filter: Selection of Sections

The filter allows to select only the samples/paths that crosses a “source node” or that crosses a “source node” and then a destination node. It is also possible to filter how many nodes/locations before the source and after the destination are desired to visualise. The remaining samples/paths are discarded.

For applying this filter, the "Activar " box must be checked.

1.3.8. Filter: Flow Disaggregation

With this filter, the corpus is divided/disaggregated into several corpus grouping the samples by their percentage of similarity. All those new corpus whose number of samples do not exceed the percentage "unir grupos menores de” (group smaller than) on the total number of initial corpus samples are unified in a same corpus. By applying this filter the initial corpus will be transformed into several corpus and put into several calculated inferences.

For applying this filter, the "Activar " box must be checked.

1.3.9. Filter: Grouped Mining

The activation of this filter means that after performing the data mining, those nodes with the same name but are separated in the workflow will be joined in one artificially. This is useful in some activities such as when locating people, because the same location can not exist in two places at once, but in other corpus this can mean distorting reality.

1.4. Data mining area

After having performed the data mining, this area allows interacting with the inference obtained. In the upper part there is a set of filters to change the data visualization or obtain views with extra data. In the lower part it is the list of corpus samples and they could be selected to visualize them.

1.4.1. List of samples

By default, it lists the corpus samples on which the mining has been applied, indicating in the upper right the number of samples. It also have controls to move through the different pages that make up the list.

In some cases the mining filters or workflow interactions will cause that not all the corpus samples be listed, but a subset; in this case, a button will appear in the left part of the list indicating that there is a filtering applied and pressing the button the filtering can be removed.

The list allows selecting a sample, green-highlighting it in the workflow, leaving semitransparent the rest of the elements. To deselect the sample, press the CTRL key while the sample is selected with the mouse.

When a sample is selected, extra data of the sample is also displayed in the Information Area, such as when it starts, when it ends, the path followed and how long it has remain in each node. (The time format is days.hours:minutes:seconds.miliseconds)

1.4.2. Mining filters

The different buttons at the top allow to chage how the inferred workflow is displayed or show extra data about the visualization.

From left to right they are:

1. Frequency map: Shows the workflow in heat map format indicating which elements occur more frequently or for a longer time.

2. Occupation map: Shows the workflow in heat map format indicating which nodes/locations are more occupied at a specific time moment.

3. Jump finder: Highlights in the workflow round-trip steps (jumps) through a node of a limited time.

4. Compare inference with a saved pattern: Allows comparing the workflow of the visualized inference with another one previously stored on disk.

5. Alpha transparency value: Allows changing the transparency value of objects that become transparent in some of the previous visualizations.

6. Restore: Restores the display to its original state.

Frequency map

It changes the visualization of the inferred workflow and shows it in a heat map format, from red, through orange and reaching green, to indicate in the case of transitions how often they occur and in the case of nodes how long they occur. Green elements have the lowest occurrence/duration while the red ones have the highest.

The filter allows displaying the frequency calculations when placing the mouse on top of a workflow element. When pressing the button "Fijar descripciones emergentes” (Set pop-up descriptions) it shows this information in a fixed way.

The sliding bars allow changing the lower and upper values of the heat map, thus forcing them to group to the maximum or minimum from a specific value.

Occupation map

The occupation map is only available if each node or localization maximum occupation data have been loaded or calculated when loading the corpus.

It allows changing the inferred workflow visualization, showing it in a heat map, from red, through orange and arriving at green, indicating which nodes/locations are more occupied in a specific moment, regarding the maximum occupation value of each node/location. The red ones are the nodes with occupancy closer to their maximum and the green ones the least occupied regarding their maximum.

With the “Inicio” and “Fin” (Start and End) date selectors, the period of time in which we want to calculate the filter is indicated and, therefore, the limits of the lower slide bar. The time values indicated in “Paso” (Step) delimit for how long the occupation of a node is calculated and, therefore, what is the step of the sliding bar. With the sliding bar can be observed the occupation from the indicated start date until the end date, with bar steps equivalent to the “Paso”. The date range of the occupation is displayed on the bar.

It is important that the End date be greater than the Start date and that the Step time be not zero or more than time between the deadlines. If this happens the filter is deactivated.

Finally, when positioning the mouse on any workflow node, the calculated occupancy data is displayed in a pop-up way. These data can be visualized permanently if the “Fijar descripciones emergentes” (Set pop-up descriptions) button is pressed.

Jump Finder

Modifies the workflow display to highlight jumps or what is the same, the go and come back from a node to the same source node, in a lower time than that indicated in the filter.

Each time the time is changed it is necessary to press the filter button again to update it.

When the filtering is applied, the samples list is also affected and only lists the samples containing jumps, instead of showing all the corpus samples used in the inference.

Compare inference with a saved pattern

When applying this type of filtering, PALIA ask to search on disk for a pattern to compare with; this will be an inference file previously saved with the button located in the main menu of the application. Once the file is read, the name of the chosen pattern is shown in the filter and the workflow is coloured according to the similarity of the compared inferences. In grey the common elements will appear, in red those that are in the pattern but not in the current inference and in green the elements that are in our inference but not in the the pattern.

Alpha transparency value

It changes the transparency value of those objects that are transparent, after applying some mining filters or after selecting a sample to see it highlighted.

1.5. Workflow graphical representation area

In this area are drawn the inferences obtained as a result of applying the data mining algorithm to the corpus obtained from the filtering.

The activities of a sample, in our case locations, are represented as a node. The nodes have a circular shape and contain the name inside. There are two special nodes added to all samples: the @Start node indicates the start of any sample/path and the @End node will collect all the sample/path endings. If no modification is applied, they will appear in green and red respectively. The normal nodes appear in grey and those that are the end of a sample/path are drawn with a discontinuous-lines edge.

The result of an activity will indicate the next destination, that is, the result generates a transition, the transitions are represented with an arrow from the origin node to the destination and with a name in the center of the line with the following format: "Origin node name = Activity Result "in our case that is equivalent to" Origin node name = Destination node name ".

The position of the workflow nodes can be changed by clicking with the mouse and dragging to the desired position.

When selecting a workflow transition, it is highlighted in purple; in the mining area, only the samples containing that transition are shown and in the information area the name of those samples is shown.

Those cases in which as a result of the filtering there are several corpus and therefore several workflow are inferred, they are displayed in the form of a stacked table, and each of the drawing areas can be resized. To identify the workflows it is added on each one a header with the name, the number of samples that perform it (occurrences) and the percentage that they are over the total samples of the original corpus. The selected workflow is the one with a blue border. For some filters and the zoom it will be the affected, while other filters affect all the workflows.

In the case of multiple visualization, by right clicking on one of the workflows, a contextual menu appears; it has three functions:

1. Compara con…(Compare with): Allows applying the filter "Comparar inferencia con un patrón guardado” (Compare inference with a saved pattern) but in this case it compares the selected workflow with another of those that are being visualized.

2. Copia Posiciones de…(Copy Positions of): Shows the list of the remaining displayed workflows and copies the positions of the nodes that match in name with the one selected, obtaining a similar graphic representation of the two workflows.

3. Fuerza Posiciones (Force Positions): Copy the positions of the nodes of the selected workflow to the other nodes that match in name, of the remaining workflows. In this way, all workflows will have a graphic representation similar to the selected one.

1.6. Information area

The information area shows different messages of the application: for example, when a corpus sample is incorrect and the application removes it, when the user selects something incorrect in some filters, the evolution of the data mining process, extra information when some elements are selected or potential application error messages.

It can be collapsed to save screen space by pressing the top left button and the showed text can be saved to a file by pressing the top right button.