Many applications in different domains generate time series data at an increasing rate. The continuous flow of emitted data may concern personal activities (e.g., through smart-meters or smart-plugs for electricity or water consumption) or professional activities (e.g., for monitoring heart activity or through the sensors installed on plants by farmers). This results in the production of large and complex data, usually in the form of time series. In recent years, there has been an explosion of interest in time series data mining. As a general rule, large time series come along with super-high dimensionality. As a consequence, it is difficult and inefficient to directly mine the raw time series without relying on dimensionality reduction. Therefore, the representation of the data is the key to efficient and effective solutions. Given this high data volumes in time series applications, or simply the need for fast response times, it is necessary to rely on alternative, shorter representations of these series, usually with loss. This incurs approximate comparisons of time series where precision is a major issue. In this thesis, we focus on the problem of segmenting time series before their transformation into symbolic representations. For this, we propose solutions to adaptively segment time series databases by adopting a variable segment size that depends on the time series distribution. These methods reduce significantly the information loss incurred by possible splittings at different steps of the representation calculation, particularly for datasets with unbalanced (non-uniform) distributions. After reviewing the state of the art, we propose three novel approaches for efficiently segmenting time series datasets by means of variable size segments. First, we propose ASAX_EN a novel approach that performs the splitting based on the representation's entropy with an approximate algorithm using a top-down strategy. Second, we propose ASAX_SSE approach that segments the time series by taking into account the sum of squared errors (SSE) with an approximate algorithm using a bottom-up strategy. This method provides high quality results in time series representation. Also, efficient algorithms for improving the execution time of our segmentation approach have been proposed. Third, we propose EASAX_SSE our segmentation method that finds the time domain division that guarantees to minimize the SSE of the representation with an exact method. Our solutions propose a lower bounding method that allows approximating the distance between the original time series based on their representations in these approaches. They have been evaluated using several real world datasets. The results illustrate that our techniques can significantly improve the time series representation quality.