Skip to end of metadata
Go to start of metadata

Overview

Timelines is a tool for analyzing statistics for entities found in texts over a period of time. This statistics is based on how frequently entities appear together in the document set. In general, documents are bound to concrete moments in time. The Timelines feature uses this additional information (metadata) about the date when each document was created. Thus it analyzes how documents and entities are distributed over a particular period and what types of co-relations exist between them.

The Timelines components can be formed either by pre-selected entities or by the most popular entities in a given document set and time frame. For each period one can further examine the underlying set of documents forming the part of the timeline.

Generally, you can calculate three types of timeline analysis:

  • Documents timelines calculates how the documents, in which the selected entities occur, are distributed over a particular period of time.
  • Selected entities timelines calculates trends for selected entities over a particular period of time.
  • Most popular entities timelines calculates trends for the most frequently mentioned entities of a given type over a particular period of time.

Creating a timeline query

In order to generate a timeline query, you have to:

  • Define the entity or entities you want to look for by entering them in the Document filter field.
  • Specify the time period over which you want to perform the query by defining the start and the end date in the Time period field.
  • Define your search criteria further by setting your options in the Based on and Type fields.
  • Configure the time units into which the time period is divided in the generated chart by using the Granularity options.

Document filter

The Document filter field consists of a search box and a text field displaying details about the current filter of the query.

You enter the entity you will be looking for in the Search for entities box. As you start typing, the auto-complete feature automatically recognizes possible names of entities and offers a list of suggestions. Selecting an entity from the list, adds it to the Document filter.


Find how the popularity of a political figure changes through time in the international news, for example Barack Obama.


Start typing "oba" and then select "Barack Obama" from the list of suggestions.

The "Barack Obama" is listed in the Selected Entity list.

You can use an unlimited number of entities in a query by adding them to the Selected entities in the Document filter.
To remove any of the selected entities, click the icon next to it.

Time period

The Time period field consists of two search boxes: From date and To date. They set the start and the end of the time period over which your query is performed.

To define the start point of the time period, go to the From date box and click the (calendar) icon. This opens a drop down menu with a calendar view of the currently selected month and year. On top of the calendar view there are two drop-down list boxes. The first one selects the month and the second – the year. The default end date is tomorrow to allow also finding the documents annotated today.

The set of possible choices for years is defined by the corpus of all documents annotated by the Latest news.

When setting the time period, you have to define the whole date (month, year, and day), even if you want to select just a different year. If you change only the year without setting a day and a month, when you close the calendar icon, your year selection will not be taken into account.


Find how the popularity of a political figure and an organization changes through time, for example Barack Obama and the Federal Reserve in the time period between January 1st, 2009 and June 30th, 2010.


Click the From date search box and select "January" from the list of months, "2009" – from the list of years, and 1st from the calendar menu respectively. After that, set the end date (June 30th, 2010) in the same way in the To date search box.

Granularity

The Granularity field gives you the option to view the set of timelines results broken into smaller time units such as days, weeks, months, quarters, or years. The time period always starts from the first day of the respective unit. The following are the recommended options:

  • Days – when the documents, in which the selected entities occur, cover a time period of less than 40 days
  • Weeks (Sun-Sat format) - when the documents, in which the selected entities occur, cover a time period of more than 40 days and less than 6 months
  • Months - when the documents, in which the selected entities occur, cover a time period of more than 6 months and less than 2 years
  • Quarters - when the documents, in which the selected entities occur, cover a time period of more than 2 and less than 5 years
  • Year - when the documents, in which the selected entities occur, cover a time period longer than 5 years

Based on

The Based on field provides two options – popularity and occurrence. The first one shows the distribution of all documents where the selected entities occur over the defined period of time, while the second option - how frequently they appear.
For example, when you want to find how the popularity of Barack Obama and the Federal Reserve changes between January 1st, 2009 and June 30th, you can view timelines for all documents in which these two entities occur within this time period.

Alternatively, you can analyze the frequency of appearance of these two entities in documents in this time frame.

Type

The Type field also offers two options. The first one is to perform your analysis based on the entities you have specified in the Document filter field. The second option is to analyze the most popular entities of an entity type over the defined period of time. When you select Most popular entities, a new field – Options – appears and enables you to make further restrictions over your query.


Analyze the trends and interrelations between the most frequently mentioned entities of the type "Person" in the news over the period between 1 December, 2008 and 31 March, 2009.


Set your time period and granularity, and choose whether the results are based on popularity or occurrence. Then select Most popular entities in the Type field.

The Timelines analysis of most popular entity calculates ups and downs in the document flow and gives a relatively realistic evaluation of the changes in the popularity of an entity.

Options

When you select the Most popular entities option from the Type field, you can define further restrictions over your query in the Options field that opens:

  • Display…topmost entities – specifies how many of the top ranking entities are displayed at the same time. You can select a number between 1 and 20 from a list box.
  • Of type --you can choose the type of entity for which popularity over a particular time frame is analyzed. You can select the type from a drop down list representing the ontology classes.

Timeline results

You can view the results in the Timelines Result screen, presented in a chart. The chart shows the fluctuation of the entities' frequency of appearance under the given restrictions.

From there you can also navigate to the documents that formed the respective peaks and drops in popularity.

Labels:
None
Enter labels to add to this page:
Please wait 
Looking for a label? Just start typing.