Antonietta Lanza

Empowering a GIS with inductive learning capabilities: the case of INGENS☆

Abstract Information given in topographic map captions or in GIS models is often insufficient to recognize interesting geographical patterns. Some prototypes of GIS have already been extended with a knowledge-base and some reasoning capabilities to support sophisticated map interpretation processes. Nevertheless, the acquisition of the necessary knowledge is still a demanding task for which machine learning techniques can be of great help. This paper presents INGENS, a prototypical GIS which integrates machine learning tools to assist users in the task of topographic map interpretation. The system can be trained to learn operational definitions of geographical objects that are not explicitly modeled in the database. INGENS has been applied to the task of Apulian map interpretation in order to discover geographic knowledge of interest to town planners.

Spatial clustering of structured objects

Clustering is a fundamental task in Spatial Data Mining where data consists of observations for a site (e.g. areal units) descriptive of one or more (spatial) primary units, possibly of different type, collected within the same site boundary. The goal is to group structured objects, i.e. data collected at different sites, such that data inside each cluster models the continuity of socio-economic or geographic environment, while separate clusters model variation over the space. Continuity is evaluated according to the spatial organization arising in data, namely discrete spatial structure, expressing the (spatial) relations between separate sites implicitly defined by their geometrical representation and positioning. Data collected within sites that are (transitively) connected in the discrete spatial structure are clustered together according to the similarity on multi-relational descriptions representing their internal structure. CORSO is a novel spatial data mining method that resorts to a multi-relational approach to learn relational spatial data and exploits the concept of neighborhood to capture relational constraints embedded in the discrete spatial structure. Relational data are expressed in a first-order formalism and similarity among structured objects is computed as degree of matching with respect to a common generalization. The application to real-world spatial data is reported.

Discovering Geographic Knowledge: The INGENS System

INGENS is a prototypical GIS which integrates machine learning tools in order to discover geographic knowledge useful for the task of topographic map interpretation. It embeds ATRE, a novel learning system that can induce recursive logic theories from a set of training examples. An application to the problem of recognizing four morphological elements in topographic maps of the Apulia region is also illustrated.

Machine learning for map interpretation: An intelligent tool for environmental planning

The design of a user interface integrating instruments for visual and textual representation and image interpretation is a relevant problem when developing an advisory system for environmental planning. Indeed, the user of the system needs a support to the interpretation of maps, that is, a tool that segments maps and automatically associates geometric regions on a map with those semantic labels useful for applying hints and advices suggested by the environmental planning system. In the article, we present the application of symbolic machine learning techniques to the interpretation of maps. Two inductive learning systems, namely, INDUBI/CSL and ATRE, have been used to complete the knowledge base of an expert system for environmental planning. The application described concerns the recognition of four environmental concepts that are relevant for environmental protection. The positive results obtained in two different experiments prove the strength of the adopted approach for the interpretation task.

Enhancing Regression Models with Spatio-temporal Indicator Additions

The task being addressed in this paper consists of trying to forecast the future value of a time series variable on a certain geographical location, based on historical data of this variable collected on both this and other locations. In general, this time series forecasting task can be performed by using machine learning models, which transform the original problem into a regression task. The target variable is the future value of the series, while the predictors are previous past values of the series up to a certain p-length time window. In this paper, we convey information on both the spatial and temporal historical data to the predictive models, with the goal of improving their forecasting ability. We build technical indicators, which are summaries of certain properties of the spatio-temporal data, grouped in the spatio-temporal clusters and use them to enhance the forecasting ability of regression models. A case study with air temperature data is presented.

An Integrated Platform for Spatial Data Mining within a GIS Environment

The strength of GIS is in providing a rich data infiastructure for combining disparate data in meaningfid ways by using a spatial arrangement (e.g., proximity). As a toolbox, a GIS allows planners to perform spatial analysis using geo-processing functions such as map overlay, connectivity measurements or thematic map coloring. Although, this makes effective the geographic visualization of individual variables, complex multi-variate dependencies are easily overlooked. The required step to take GIS beyond a tool for automating cartography is to incorporate the ability of analyzing and condensing a large number of geo-referenced variables into a single forecast or score. This is where data mining promises great potential benefits and the reason why there is such a hand-in-glove fit between GIS and data mining. Following the mainstream of this research, we propose to integrate GIS and data mining fitnctionality in a closely coupled open and extensible GIS architecture. This is done by resorting to emerging spatial data mining technology that deals with the substantial complexity added from the spatial dimension. We illustrate an example of topographic map interpretation where resorting to data mining facilities to discover both operational definitions of morphologies characterizing the landscape (i.e., spatial classification rules) and frequent spatial interactions of two or more spatially-referred objects (i.e.. spatial association rules). In both cases, discovered patterns correspond to what geographers, geologists and town planners are interested in while interpreting a map, although they are never explicitly represented in topographic maps or in a GIS-model.

Wind Power Forecasting Using Time Series Cluster Analysis

The growing integration of wind turbines into the power grid can only be balanced with precise forecasts of upcoming energy productions. This information plays as basis for operation and management strategies for a reliable and economical integration into the power grid. A precise forecast needs to overcome problems of variable energy production caused by fluctuating weather conditions. In this paper, we define a data mining approach, in order to process a past set of the wind power measurements of a wind turbine and extract a robust prediction model. We resort to a time series clustering algorithm, in order to extract a compact, informative representation of the time series of wind power measurements in the past set. We use cluster prototypes for predicting upcoming wind powers of the turbine. We illustrate a case study with real data collected from a wind turbine installed in the Apulia region.

Geographic Knowledge Discovery in INGENS: An Inductive Database Perspective

INGENS is a prototype of GIS which integrates a geographic knowledge discovery engine to mine several kinds of spatial KDD objects from the topographic maps stored in a spatial database. In this paper we describe the main principles of an inductive spatial database in INGENS. Inductive database allows to keep permanent KDD objects and integrate database technology with systems for the geographic knowledge generation. In contrast to traditional spatial database technology, inductive database allows to answer queries which require synthesizing and applying plausible knowledge which is generated by (inductive) inference from both spatial objects and KDD objects (prior knowledge) stored in the same database.

Generating Logic Descriptions for the Automated Interpretation of Topographic Maps

Automating the interpretation of a map in order to locate some geographical objects and their relations is a challenging task, which goes beyond the transformation of map images into a vectorized representation and the recognition of symbols. In this work, we present an approach to the automated interpretation of vectorized topographic maps. It is based on the generation of logic descriptions of maps and the application of symbolic Machine Learning tools to these descriptions. This paper focuses on the definition of computational methods for the generation of logic descriptions of map cells and briefly describes the use of these logic descriptions in an inductive learning task.

Very Short-Term Wind Speed Forecasting Using Spatio-Temporal Lazy Learning

A wind speed forecast corresponds to an estimate of the upcoming production of a wind farm. The paper illustrates a variant of the Nearest Neighbor algorithm that yields wind speed forecasts, with a fast time resolution, for a (very) short time horizon. The proposed algorithm allows us to monitor a grid of wind farms, which collaborate by sharing information (i.e. wind speed measurements). It accounts for both spatial and temporal correlation of shared information. Experiments show that the presented algorithm is able to determine more accurate forecasts than a state-of-art statistical algorithm, namely auto. ARIMA.