Búsqueda avanzada

El autor Susana Ladra ha publicado 5 artículo(s):

1 - Scalable and queryable compressed storage structure for raster data

Titulo: Scalable and queryable compressed storage structure for raster data Autores: Susana Ladra, José R. Paramá, Fernando Silva-Coira Revista: Information Systems Volume 72, December 2017, Pages 179-204Factor de impacto: 2.777Ranking JCR: Q2Citas: 1 DOI:

Autores: Susana Ladra / Jose R. Parama / Fernando Silva-Coira / 
Palabras Clave: Compresión de datos - Ráster - Sistemas de Informacíon Geográfica

2 - Compact and queryable representation of raster datasets

Titulo: Compact and queryable representation of raster datasets Autores: Susana Ladra, José R. Paramá, Fernando Silva-Coira Congreso: INTERNATIONAL CONFERENCE ON SCIENTIFIC AND STATISTICAL DATA BASE MANAGEMENT (SSDBM) 2016 Clasificación Ranking SCIE. Clase 2 (A-) Clasidicación CORE: A Citas: 2 DOI:

Autores: Susana Ladra / José R. Paramá / Fernando Silva-Coira / 
Palabras Clave: Compresión de datos - Ráster - Sistemas de Informacíon Geográfica

3 - Exploiting SIMD instructions in current processors to improve classical string algorithms

Algorithms and data structures for efficient representation and processing of large databases can be combined with advances in computer architecture, as hardware-aware implementations that exploit particular hardware features. For example, many algorithms have been adapted to exploit the architecture of GPUs, FPGAs, or general-purpose CPUs providing instructions included for particular application domains.
In this paper we explore how the Intel SSE4.2 (Streaming SIMD Extensions) SIMD (Single Instruction Multiple Data) instructions included in Intel/AMD processors can improve the performance of algorithms for text indexing and searching. The SSE4.2 instruction subset provides instructions for text processing. Our implementations are mainly based on the followings: POPCOUNT counts the number of 1 bits in a word of up to 64 bits, PCMPESTRI compares two strings of length up to 16 bytes and returns the result as a binary mask.
Despite the benefits these features can bring to text processing, they have been rarely used or evaluated in the existing literature. We present case studies and experimental results that show how much text/string algorithms can benefit from the SIMD extensions. Particularly, we focus on the rank and select operations in sequences of bits and bytes, and the Horspool string search algorithm.

Autores: Susana Ladra / Oscar Pedreira / Jose Duato / Nieves R. Brisaboa / 
Palabras Clave:

4 - A Compact Representation of Indoor Trajectories

We present a system that combines indoor positioning with a compression algorithm for trajectories in the context of a nursing home. Our aim is to gather and effectively represent the location of the residents and caregivers along time, while allowing for efficient access to those data.We briefly show the system architecture that enables the automatic tracking of user’s movements and consequently the gathering of their locations. Then, we present indRep, our compact representation to handle positioning data using grammar-based compression, and provide two basic operations that enable pseudo-random access to the data. Finally, we include experiments that show that indRep is competitive with well-know general-purpose compressors in terms of compression effectiveness and also provides fast access to the compressed data. We expect both features would enable exploitation functionalities even in computers with rather low computational resources.

Autores: Antonio Fariña / Pablo Gutiérrez-Asorey / Susana Ladra / Miguel R. Penabad / Tirso V. Rodeiro / 
Palabras Clave: Compression - Data Structures - Indoor Trajectories

5 - Practical Representations for Web and Social Graphs

Graphs are a natural way of modeling connections among Web pages in a network or people according to a criteria like friendship, co-authorship, etc. Many algorithms that compute and infer interesting facts out of these networks work directly over these graphs. Some examples of this are Connected components, HITS, PageRank, spamdetection, among others. In social networks, graph mining also enables the study of populations behavior. Successful graph mining not only enables segmentation of users but also prediction of behavior. Link analysis and graph mining remains an area of high interest and development.
These human-generated graphs are growing at an amazing pace, and their representation in main memory, secondary memory, and distributed systems are getting more and more attention. Furthermore, space-efficient representations for these graphs have succeeded at exploiting regularities that are particular to the domain. In the case of Web graphs the main properties exploited are the locality of reference, skewed in/out-degree, and similarity of adjacency lists among nodes of the same domain. In social networks, there is a tendency towards forming cliques in the network.

Autores: Francisco Claude / Susana Ladra / 
Palabras Clave: