TY - CPAPER TI - Scalable summary based retrieval in P2P networks AU - Müller, Wolfgang AU - Eisenhardt, Martin AU - Henrich, Andreas PY - 2025 SP - 586-593 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/106621 LA - deu AB - Much of the present P2P-IR literature is focused on distributed indexing structures. Within this paper, we present an approach based on the replication of peer data summaries via rumor spreading and multicast in a structured overlay.We will describe Rumorama, a P2P framework for similar-ity queries inspired by GlOSS and CORI and their P2P-adaptation, PlanetP. Rumorama achieves a hierarchization of PlanetP-like summary-based P2P-IR networks. In a Rumorama network, each peer views the network as a small PlanetP network with connections to peers that see other small PlanetP networks. One important aspect is that each peer can choose the size of the PlanetP network it wants to see according to its local processing power and bandwidth. Even in this adaptive environment, Rumorama manages to process a query such that the summary of each peer is considered exactly once in a network without churn. However, the actual number of peers to be contacted for a query is a small fraction of the total number of peers in the network.Within this article, we present the Rumorama base protocol, as well as experiments demonstrating the scalability and viability of the approach under churn. KW - P2P KW - Similarity Search KW - Distributed IR ER - TY - CPAPER TI - Sample-based Creation of Peer Summaries for Efficient Similarity Search in Scalable Peer-to-Peer Networks AU - Blank, Daniel AU - El Allali, Soufyane AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2025 SP - 143-151 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/106620 LA - eng AB - In this paper we introduce a simple yet experimentally convincing approach in the research field of source selection for content-based similarity search in P2P networks or, more concretely, in summary-based P2P systems. In these systems, summaries are used for data source selection when performing k-NN queries on distributed collections of documents represented by feature vectors. We introduce a new type of cluster-based summaries for source selection that can efficiently and cheaply be calculated and distributed in P2P networks. For the summaries generation, a very large number of sample points is used. Each peer in the network assigns its indexing data to their corresponding closest sample points and publishes its constructed summary. We evaluate the quality of these summaries when changing the number of sample points used in experiments on real-world image feature data obtained from a large crawl of the flickr web photo community and show that for higher numbers of sample points we achieve a better retrieval performance. Our experiments show that the proposed summaries yield four times better performance with respect to previous methods. Intuitively, there are some disadvantages to this approach due to the large size of the generated summaries. We show experimentally, that these disadvantages can easily be overcome due to the sparse nature of the generated summaries by simple compression techniques. KW - P2P KW - CBIR KW - Source selection ER - TY - CPAPER TI - Fast Retrieval of High-Dimensional Feature Vectors in P2P Networks Using Compact Peer Data Summaries AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2025 SP - 79–86 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/106613 LA - deu AB - The retrieval facilities of most Peer-to-Peer (P2P) systems are limited to queries based on a unique identifier or a small set of keywords. The techniques used for this purpose are hardly applicable for content-based image retrieval (CBIR) in a P2P network. Furthermore, we will argue that the curse of dimensionality and the high communication overhead prevent the adaptation of multidimensional search trees or fast sequential scan techniques for P2P CBIR. In the present paper we will propose two compact data representations which can be distributed in a P2P network and used as the basis for a source selection. This allows to communicate only with a small fraction of all peers during query processing without deteriorating the result quality significantly. We will also present experimental results confirming our approach. KW - Algorithms KW - Performance ER - TY - CPAPER TI - Aspects of Adaptivity in P2P Information Retrieval AU - Müller, Wolfgang AU - Henrich, Andreas AU - Eisenhardt, Martin PY - 2025 SP - 164-178 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/106380 LA - eng AB - Peer-to-Peer networks are comprised of multiple independently administered computers (peers) that cooperate via a common protocol in order to achieve a goal common to the peers. Helping the user find relevant information in a P2P network is the subject of the field of Peer-to-Peer IR. In order to be successful, a P2P-IR system needs to be adaptive in several respects. It has to adapt both to the user and to its environment. Within this article we detail the motivations and challenges of P2P-IR, as well as the ways in which P2P-IR systems adapt and where improvement is needed in order to achieve adaptive multimedia retrieval. KW - P2P Information Retrieval ER - TY - CPAPER TI - Image Data Source Selection Using Gaussian Mixture Models AU - El Allali, Soufyane AU - Blank, Daniel AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2025 SP - 170-181 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/106376 LA - eng AB - In peer-to-peer (P2P) networks, computers with equal rights form a logical (overlay) network in order to provide a common service that lies beyond the capacity of every single participant. Efficient similarity search is generally recognized as a frontier in research about P2P systems. In literature, a variety of approaches exist. One of which is data source selection based approaches where peers summarize the data they contribute to the network, generating typically one summary per peer. When processing queries, these summaries are used to choose the peers (data sources) that are most likely to contribute to the query result. Only those data sources are contacted. In this paper we use a Gaussian mixture model to generate peer summaries using the peers’ local data. We compare this method to other local unsupervised clustering methods for generating peer summaries and show that a Gaussian mixture model is promising when it comes to locally generated summaries for peers without the need for a distributed summary computation that needs coordination between peers. KW - Image Data KW - Source Selection KW - Gaussian Mixture Models ER - TY - CPAPER TI - Clustering-Based Source Selection for Efficient Image Retrieval in Peer-to-Peer Networks AU - Eisenhardt, Martin AU - Müller, Wolfgang AU - Henrich, Andreas AU - Blank, Daniel AU - El Allali, Soufyane PY - 2025 SP - 1-6 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/106883 LA - deu AB - In peer-to-peer (P2P) networks, computers with equal rights form a logical (overlay) network in order to provide a common service that lies beyond the capacity of every single participant. Efficient similarity search is generally recognized as a frontier in research about P2P systems. One way to address it is using data source selection based approaches where peers summarize the data they contribute to the network, generating typically one summary per peer. When processing queries, these summaries are used to choose the peers (data sources) that are most likely to contribute to the query result. Only those data sources are contacted. There are two main contributions of this paper. We extend earlier work, adding a data source selection method for high-dimensional vector data, comparing different peer ranking schemes. More importantly, we present a method that uses progressive stepwise data exchange between peers to better each peer's summary and therefore improve the system's performance. KW - Efficient Image Retrieval KW - Peer-to-Peer Networks ER - TY - CPAPER TI - Picadomo : Faceted Image Browsing for Mobile Devices AU - Hub, Adrian AU - Blank, Daniel AU - Henrich, Andreas AU - Müller, Wolfgang PY - 2025 SP - 249-254 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/106877 LA - eng AB - Picadomo combines content-based image retrieval and faceted search on mobile devices. It is designed for finding images with desired visual properties, tags or other known metadata. Due to the limitations of mobile devices such as small screen sizes and low processing power, we had to carefully select the features that come in use (dominant color, GPS data, tags, etc.). With Picadomo the user can pick visualized facets directly via touch screen, while using very little screen size for the facet browsing navigation. We present our architecture, the facets used for image browsing, our new control concept and user experiments. KW - Picadomo ER - TY - CPAPER TI - VisualFlamenco : Dependable, Interactive Image Browsing Based on Visual Properties AU - Müller, Wolfgang AU - Zech, Markus AU - Henrich, Andreas AU - Blank, Daniel PY - 2025 SP - 568-575 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/106875 LA - eng AB - Faceted search as a method for exploratory search is a search interface technique for Boolean retrieval that provides a good compromise between the wish to guide the user and the need to give freedom to the user. Both, guidance and freedom are needed in order to enable him (or her) to accomplish his search goal as quickly and efficiently as possible. The search goal can be both, to retrieve a target image (lookup search) or to get an overview of a collection as the basis for accomplishing complex search tasks (exploratory search). VisualFlamenco is about bringing content-based image retrieval and faceted search together. While there is some preexisting work towards faceted search based on automatically extracted metadata, the base innovation of our work is to visualize the meaning of facets and let the user pick visualized facets directly. The main point is this research is to give the user the right expectations about what the system can achieve. We see this as a step towards dependable multimedia search, search interfaces that do not promise more than they can achieve. A mismatch of expectations and possibilities has been denoted by some as one of the main problems in current research about multimedia search. Within this paper we present a visual faceted search tool that seeks to provide dependable image search. We describe its base idea, the features used, as well as user experiments done using the system. KW - VisualFlamenco ER - TY - CPAPER TI - Faster exact histogram intersection on large data collections using inverted VA-files AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2025 SP - 455-463 PB - Otto-Friedrich-Universität CY - Bamberg T3 - Lecture notes in computer science UR - https://fis.uni-bamberg.de/handle/uniba/106897 LA - eng AB - Most indexing structures for high-dimensional vectors used in multimedia retrieval today rely on determining the importance of each vector component at indexing time in order to create the index. However for Histogram Intersection and other important distance measures this is not possible because the importance of vector components depends on the query. We present an indexing structure inspired by VA-file and Inverted file that does not need to determine the importance at indexing time in order to perform well. Instead, our structure adapts to the importance of vector components at query processing time. Success of this approach is demonstrated in experiments on feature data extracted from a large image collection. KW - histogram intersection KW - large data collections ER - TY - JOUR TI - Inhaltsbasierte Suche in P2P-Systemen AU - Eisenhardt, Martin AU - Henrich, Andreas AU - Müller, Wolfgang PY - 2025 SP - 5-15 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/106989 LA - deu AB - Pcer-to-Peer-Netze ( P2P-Netze) erfreuen sich zunehmender Beliebtheit hei der verteilten und administrationsarmen Verwaltung größerer Dokumentenkollektionen. Ein wichtiges Problem ist dabei die Suche und insbesondere die inhaltsbasierte Suche nach Dokumenten in P2P-Netzen. Der vorliegende Beitrag gibt einen Überblick über entsprechende Ansätze und zeigt interessante Forschungsaspekte auf. KW - P2P-Systeme ER - TY - CPAPER TI - Comparing the performance of two CBIRS indexing schemes AU - Müller, Wolfgang AU - Robbert, Günter AU - Henrich, Andreas PY - 2025 SP - 9-20 PB - Otto-Friedrich-Universität CY - Bamberg SN - 0277-786X UR - https://fis.uni-bamberg.de/handle/uniba/106994 LA - eng AB - Content based image retrieval (CBIR) as it is known today has to deal with a number of challenges. Quickly summarized, the main challenges are firstly, to bridge the semantic gap between high-level concepts and low-level features using feedback, secondly to provide performance under adverse conditions. High-dimensional spaces, as well as a demanding machine learning task make the right way of indexing an important issue. When indexing multimedia data, most groups opt for extraction of high-dimensional feature vectors from the data, followed by dimensionality reduction like PCA (Principal Components Analysis) or LSI (Latent Semantic Indexing). The resulting vectors are indexed using spatial indexing structures such as kd-trees or R-trees, for example. Other projects, such as MARS and Viper propose the adaptation of text indexing techniques, notably the inverted file. Here, the Viper system is the most direct adaptation of text retrieval techniques to quantized vectors. However, while the Viper query engine provides decent performance together with impressive user-feedback behavior, as well as the possibility for easy integration of long-term learning algorithms, and support for potentially infinite feature vectors, there has been no comparison of vector-based methods and inverted-file-based methods under similar conditions. In this publication, we compare a CBIR query engine that uses inverted files (Bothrops, a rewrite of the Viper query engine based on a relational database), and a CBIR query engine based on LSD (Local Split Decision) trees for spatial indexing using the same feature sets. The Benchathlon initiative works on providing a set of images and ground truth for simulating image queries by example and corresponding user feedback. ·when performing the Benchathlon benchmark on a CBIR system (the System Under Test, SUT), a benchmarking harness connects over internet to the SUT, performing a number of queries using an agreed-upon protocol, the multimedia retrieval markup language (MRML). Using this bench-mark one can measure the quality of retrieval, as well as the overall (speed) performance of the benchmarked system. Our Benchmarks will draw on the Benchathlon's work for documenting the retrieval performance of both inverted file-based and LSD tree based techniques. However, in addition to these results, we will present statistics, that can be obtained only inside the system under test. These statistics will include the number of complex mathematical operations, as well as the amount of data that has to be read from disk during operation of a query. KW - CBIRS indexing schemes ER - TY - CPAPER TI - Efficient content-based P2P image retrieval using peer content descriptions AU - Müller, Wolfgang AU - Eisenhardt, Martin AU - Henrich, Andreas PY - 2025 SP - 57-68 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/107002 LA - eng AB - Peer-to-peer (P2P) networks are overlay networks that connect independent computers (also called nodes or peers). In contrast to client/server solutions, all nodes offer and request services from other peers in a P2P network. P2P networks are very attractive in that they harness the computing power of many common desktop machines and necessitate little administrative overhead. While the resulting computing power is impressive, efficiently looking up data still is the major challenge in P2P networks. Current work comprises fast lookup of one-dimensional values (Distributed Hash Tables, DHT) and retrieval of texts using few keywords. However, the lookup of multimedia data in P2P networks is still attacked by very few groups. In this paper, we present experiments with efficient Content Based Image Retrieval in a P2P environment, thus a P2P-CBIR system. The challenge in such systems is to limit the number of messages sent, and to maximize the usefulness of each peer contacted in the query process. We achieve this by distributing peer data summaries over the network. Obviously, the data summaries have to be compact in order to limit the communication overhead. We propose an CBIR scheme based on a compact peer data summary. This peer data summary relies on cluster frequencies. To obtain the compact representation of a peer's collection, a. global clustering of the data is efficiently calculated in a distributed manner. After that, each peer publishes how many of its images fall into each cluster. These cluster frequencies are then used by the querying peer to contact only those peers that have the largest number of images present in one cluster given by the query. In our paper we further detail the various challenges that have to be met by the designers of such a P2P- CBIR, and we present experiments with varying degree of data replication (duplicates of images) , as well as quality of clustering within the network. KW - P2P ER - TY - CPAPER TI - Parameters in P2P-IR AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2023 SP - 4-11 PB - Otto-Friedrich-Universität CY - Bamberg T3 - GI-Edition : Proceedings UR - https://fis.uni-bamberg.de/handle/uniba/92263 LA - eng AB - P2P Information Retrieval (P2P-IR) is a challenging and potentially very useful area of research. Its goal is to enable powerful, efficient, massively distributed IR. A large fraction of current research is focused on efficiency. However, we feel that P2P-IR needs to address the complicated combination of indexing structure and networking issues. KW - Parameter ER - TY - CPAPER TI - Spektrale Bloom-Filter für Peer-to-Peer Information Retrieval AU - Eisenhardt, Martin AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2023 SP - 44-48 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/92268 LA - deu AB - Spektrale Bloomfilter können als Zusammenfassung von Peer-Daten in verteiltem P2P-Retrieval verwendet werden. Dies ermöglicht eine messbar bessere Auswahl derjenigen Peers, an die eine Anfrage gestellt wird, und somit auch eine höhere Effizienz der Anfragebearbeitung. Experimentelle Daten belegen diese Verbesserung des Retrievals in P2P-Netzen. KW - Spektrale Bloom-Filter ER - TY - CPAPER TI - Untersuchung des Einflusses verschiedener Bild- Features und Distanzmaße im inhaltsbasierten P2P Information Retrieval AU - El Allali, Soufyane AU - Blank, Daniel AU - Eisenhardt, Martin AU - Henrich, Andreas AU - Müller, Wolfgang PY - 2023 SP - 382-396 PB - Otto-Friedrich-Universität CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/92293 LA - deu AB - Gegenstand der vorliegenden Arbeit ist die feature-basierte Ähnlichkeitssuche in Bilddatenbeständen, die in einem P2P-Netz verteilt sind. Wir untersuchen dabei die Eignung verschiedener Featuresätze und Distanzmaße für die Nutzung in diesem Szenario. Hierzu beziehen wir uns primär auf PlanetP-artige P2P-Netze und vergleichen die in Abhängigkeit von der Anzahl der kontaktierten Peers erreichten Ergebnisse zunächst mit einem zentralen System mit gleichem Featuresatz und Distanzmaß. Ferner vergleichen wir unser System mit einer Erweiterung, die Indexdaten im P2P-Netz transferiert, sowie mit der Implementierung einer auf CANs basierenden verteilten Indexstruktur. Schließlich evaluieren wir das System auch mittels Relevanzbeurteilungen, die von Testnutzern gegeben wurden. Die Ergebnisse zeigen deutlich die unterschiedliche Eignung verschiedener Featuresätze und Distanzmaße für unser P2P-Szenario auf. KW - Information Retrieval ER - TY - JOUR TI - Aiming Higher : The Consequences of Progressive Ambition among MPs in European Parliaments AU - Sieberer, Ulrich AU - Müller, Wolfgang PY - 2017 JO - European political science review : EPSR ; a journal of the European Consortium for Political Research VL - 9 IS - 1 SP - 27-50 PB - Cambridge Univ. Press CY - Cambridge SN - 1755-7739 DO - 10.1017/S1755773915000260 UR - https://fis.uni-bamberg.de/handle/uniba/40480 LA - eng ER - TY - CPAPER TI - Picadomo : Faceted Image Browsing for Mobile Devices AU - Hub, Adrian AU - Blank, Daniel AU - Henrich, Andreas AU - Müller, Wolfgang PY - 2009 T2 - 2009 Seventh International Workshop on Content-Based Multimedia Indexing SP - 249-254 PB - IEEE CY - Piscataway, NJ DO - 10.1109/CBMI.2009.34 UR - https://fis.uni-bamberg.de/handle/uniba/18766 LA - eng AB - Picadomo combines content-based image retrieval and faceted search on mobile devices. It is designed for finding images with desired visual properties, tags or other known metadata. Due to the limitations of mobile devices such as small screen sizes and low processing power, we had to carefully select the features that come in use (dominant color, GPS data, tags, etc.). With Picadomo the user can pick visualized facets directly via touch screen, while using very little screen size for the facet browsing navigation. We present our architecture, the facets used for image browsing, our new control concept and user experiments. KW - Picadomo ER - TY - CPAPER TI - Designing Benchmarks for the Evaluation of Peer-to- Peer Information Retrieval Systems AU - Blank, Daniel AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2008 T2 - LWA 2008 : Workshop-Woche: Lernen, Wissen & Adaptivität ; Würzburg, 6.-8. Oktober 2008 SP - 57 - 64 CY - Würzburg T3 - University of Würzburg, Germany / Department of Computer Science: Technical Report UR - https://fis.uni-bamberg.de/handle/uniba/17551 LA - eng ER - TY - CPAPER TI - Image Data Source Selection Using Gaussian Mixture Models AU - El Allali, Soufyane AU - Blank, Daniel AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2008 T2 - Adaptive multimedial retrieval: retrieval, user, and semantics : 5th International Workshop, AMR 2007, Paris, France, July 5 - 6, 2007 ; revised selected papers SP - 170-181 PB - Springer CY - Berlin u.a. SN - 978-3-540-79859-0 DO - 10.1007/978-3-540-79860-6_14 UR - https://fis.uni-bamberg.de/handle/uniba/17555 LA - eng AB - In peer-to-peer (P2P) networks, computers with equal rights form a logical (overlay) network in order to provide a common service that lies beyond the capacity of every single participant. Efficient similarity search is generally recognized as a frontier in research about P2P systems. In literature, a variety of approaches exist. One of which is data source selection based approaches where peers summarize the data they contribute to the network, generating typically one summary per peer. When processing queries, these summaries are used to choose the peers (data sources) that are most likely to contribute to the query result. Only those data sources are contacted. In this paper we use a Gaussian mixture model to generate peer summaries using the peers’ local data. We compare this method to other local unsupervised clustering methods for generating peer summaries and show that a Gaussian mixture model is promising when it comes to locally generated summaries for peers without the need for a distributed summary computation that needs coordination between peers. KW - Image Data KW - Source Selection KW - Gaussian Mixture Models ER - TY - CPAPER TI - VisualFlamenco : Dependable, Interactive Image Browsing Based on Visual Properties AU - Müller, Wolfgang AU - Zech, Markus AU - Henrich, Andreas AU - Blank, Daniel PY - 2008 T2 - Proceedings: 2008 International Workshop on Content-Based Multimedia Indexing : CBMI 2008 ; London, United Kingdom, 18 - 20 June 2008 SP - 568-575 PB - IEEE CY - Piscataway, NJ DO - 10.1109/CBMI.2008.4564998 UR - https://fis.uni-bamberg.de/handle/uniba/17554 LA - eng AB - Faceted search as a method for exploratory search is a search interface technique for Boolean retrieval that provides a good compromise between the wish to guide the user and the need to give freedom to the user. Both, guidance and freedom are needed in order to enable him (or her) to accomplish his search goal as quickly and efficiently as possible. The search goal can be both, to retrieve a target image (lookup search) or to get an overview of a collection as the basis for accomplishing complex search tasks (exploratory search). VisualFlamenco is about bringing content-based image retrieval and faceted search together. While there is some preexisting work towards faceted search based on automatically extracted metadata, the base innovation of our work is to visualize the meaning of facets and let the user pick visualized facets directly. The main point is this research is to give the user the right expectations about what the system can achieve. We see this as a step towards dependable multimedia search, search interfaces that do not promise more than they can achieve. A mismatch of expectations and possibilities has been denoted by some as one of the main problems in current research about multimedia search. Within this paper we present a visual faceted search tool that seeks to provide dependable image search. We describe its base idea, the features used, as well as user experiments done using the system. KW - VisualFlamenco ER - TY - JOUR TI - Comparison of Image Similarity Queries in P2P Systems AU - Müller, Wolfgang AU - Boykin, P. O. AU - Sarshar, N. AU - Roychowdhury, V. P. PY - 2008 JO - Computer communications VL - 31 IS - 2, 5 February 2008 = Special Issue: Foundation of Peer-to-Peer Computing SP - 375 - 386 SN - 0140-3664 UR - https://fis.uni-bamberg.de/handle/uniba/16360 LA - deu ER - TY - CPAPER TI - Comparison of Image Similarity Queries in P2P Systems AU - Müller, Wolfgang AU - Boykin, P. O. AU - Sarshar, N. AU - Roychowdhury, V. P. PY - 2007 T2 - International Conference on Peer-to-Peer Computing <7, 2007, Galway>: Proceedings SP - 98 - 105 CY - Los Alamitos, Calif. u.a. UR - https://fis.uni-bamberg.de/handle/uniba/16358 LA - eng ER - TY - CPAPER TI - Sample-based Creation of Peer Summaries for Efficient Similarity Search in Scalable Peer-to-Peer Networks AU - Blank, Daniel AU - El Allali, Soufyane AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2007 T2 - MIR '07 : Proceedings of the international workshop on Workshop on multimedia information retrieval SP - 143-151 PB - ACM CY - New York SN - 978-1-59593-778-0 DO - 10.1145/1290082.1290104 UR - https://fis.uni-bamberg.de/handle/uniba/16361 LA - eng AB - In this paper we introduce a simple yet experimentally convincing approach in the research field of source selection for content-based similarity search in P2P networks or, more concretely, in summary-based P2P systems. In these systems, summaries are used for data source selection when performing k-NN queries on distributed collections of documents represented by feature vectors. We introduce a new type of cluster-based summaries for source selection that can efficiently and cheaply be calculated and distributed in P2P networks. For the summaries generation, a very large number of sample points is used. Each peer in the network assigns its indexing data to their corresponding closest sample points and publishes its constructed summary. We evaluate the quality of these summaries when changing the number of sample points used in experiments on real-world image feature data obtained from a large crawl of the flickr web photo community and show that for higher numbers of sample points we achieve a better retrieval performance. Our experiments show that the proposed summaries yield four times better performance with respect to previous methods. Intuitively, there are some disadvantages to this approach due to the large size of the generated summaries. We show experimentally, that these disadvantages can easily be overcome due to the sparse nature of the generated summaries by simple compression techniques. KW - P2P KW - CBIR KW - Source selection ER - TY - CPAPER TI - Untersuchung des Einflusses verschiedener Bild- Features und Distanzmaße im inhaltsbasierten P2P Information Retrieval AU - El Allali, Soufyane AU - Blank, Daniel AU - Eisenhardt, Martin AU - Henrich, Andreas AU - Müller, Wolfgang PY - 2007 T2 - Datenbanksysteme in Business, Technologie und Web (BTW) : 12. Fachtagung des GI-Fachbereichs "Datenbanken und Informationssysteme" (DBIS) ; 07. - 09.03.2007 in Aachen SP - 382 - 396 PB - Gesellschaft für Informatik e.V. CY - Bonn SN - 978-3-88579-197-3 UR - https://fis.uni-bamberg.de/handle/uniba/16364 LA - deu AB - Gegenstand der vorliegenden Arbeit ist die feature-basierte Ähnlichkeitssuche in Bilddatenbeständen, die in einem P2P-Netz verteilt sind. Wir untersuchen dabei die Eignung verschiedener Featuresätze und Distanzmaße für die Nutzung in diesem Szenario. Hierzu beziehen wir uns primär auf PlanetP-artige P2P-Netze und vergleichen die in Abhängigkeit von der Anzahl der kontaktierten Peers erreichten Ergebnisse zunächst mit einem zentralen System mit gleichem Featuresatz und Distanzmaß. Ferner vergleichen wir unser System mit einer Erweiterung, die Indexdaten im P2P-Netz transferiert, sowie mit der Implementierung einer auf CANs basierenden verteilten Indexstruktur. Schließlich evaluieren wir das System auch mittels Relevanzbeurteilungen, die von Testnutzern gegeben wurden. Die Ergebnisse zeigen deutlich die unterschiedliche Eignung verschiedener Featuresätze und Distanzmaße für unser P2P-Szenario auf. KW - Information Retrieval ER - TY - CPAPER TI - Aspects of Adaptivity in P2P Information Retrieval AU - Müller, Wolfgang AU - Henrich, Andreas AU - Eisenhardt, Martin PY - 2007 T2 - Adaptive Multimedia Retrieval:User, Context, and Feedback : 4th International Workshop, AMR 2006, Geneva, Switzerland, July, 27-28, 2006 ; Revised Selected Papers SP - 164-178 PB - Springer CY - Berlin ; Heidelberg SN - 978-3-540-71544-3 DO - 10.1007/978-3-540-71545-0_13 UR - https://fis.uni-bamberg.de/handle/uniba/14573 LA - eng AB - Peer-to-Peer networks are comprised of multiple independently administered computers (peers) that cooperate via a common protocol in order to achieve a goal common to the peers. Helping the user find relevant information in a P2P network is the subject of the field of Peer-to-Peer IR. In order to be successful, a P2P-IR system needs to be adaptive in several respects. It has to adapt both to the user and to its environment. Within this article we detail the motivations and challenges of P2P-IR, as well as the ways in which P2P-IR systems adapt and where improvement is needed in order to achieve adaptive multimedia retrieval. KW - P2P Information Retrieval ER - TY - CPAPER TI - Clustering-Based Source Selection for Efficient Image Retrieval in Peer-to-Peer Networks AU - Eisenhardt, Martin AU - Müller, Wolfgang AU - Henrich, Andreas AU - Blank, Daniel AU - El Allali, Soufyane PY - 2006 T2 - Proceedings : ISM 2006, Eighth IEEE International Symposium on Multimedia : 11 - 13 December 2006, San Diego, CA SP - 823-830 PB - IEEE CY - Los Alamitos, Calif. DO - 10.1109/ISM.2006.47 UR - https://fis.uni-bamberg.de/handle/uniba/16359 LA - deu AB - In peer-to-peer (P2P) networks, computers with equal rights form a logical (overlay) network in order to provide a common service that lies beyond the capacity of every single participant. Efficient similarity search is generally recognized as a frontier in research about P2P systems. One way to address it is using data source selection based approaches where peers summarize the data they contribute to the network, generating typically one summary per peer. When processing queries, these summaries are used to choose the peers (data sources) that are most likely to contribute to the query result. Only those data sources are contacted. There are two main contributions of this paper. We extend earlier work, adding a data source selection method for high-dimensional vector data, comparing different peer ranking schemes. More importantly, we present a method that uses progressive stepwise data exchange between peers to better each peer's summary and therefore improve the system's performance. KW - Efficient Image Retrieval KW - Peer-to-Peer Networks ER - TY - CPAPER TI - Parameters in P2P-IR AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2006 T2 - Informatik 2006: Informatik für Menschen ; Beiträge der 36. Jahrestagung der Gesellschaft für Informatik e.V. (GI) ; 2. bis 6. Oktober 2006 in Dresden SP - 4-11 PB - Gesellschaft für Informatik e.V. CY - Bonn T3 - GI-Edition : Proceedings SN - 978-3-88579-187 UR - https://fis.uni-bamberg.de/handle/uniba/14818 LA - eng AB - P2P Information Retrieval (P2P-IR) is a challenging and potentially very useful area of research. Its goal is to enable powerful, efficient, massively distributed IR. A large fraction of current research is focused on efficiency. However, we feel that P2P-IR needs to address the complicated combination of indexing structure and networking issues. KW - Parameter ER - TY - JOUR TI - Anonymität und Sicherheit in Peer-to-Peer-Systemen AU - Müller, Wolfgang PY - 2006 JO - Datenbank-Spektrum : Zeitschrift für Datenbanktechnologie ; Organ der Fachgruppe Datenbanken der Gesellschaft für Informatik e.V. VL - 6 IS - 16 = Themenschwerpunkt: Security & Privacy SP - 15-25 SN - 1618-2162 UR - https://fis.uni-bamberg.de/handle/uniba/14618 LA - deu ER - TY - CPAPER TI - Robustes, skalierbares GlOSS in P2P-Netzen mit Rumorama AU - Müller, Wolfgang AU - Eisenhardt, Martin PY - 2005 T2 - WebDB Meets IR : proceedings ; Workshop des GI- Arbeitskreises "Web und Datenbanken" und der GI-Fachgruppe "Information Retrieval" im Rahmen der 11. GI-Fachtagung für Datenbanksysteme in Business, Technologie und Web (BTW 2005), Karlsruhe, 1. März 2005 SP - 45 - 54 CY - Karlsruhe UR - https://fis.uni-bamberg.de/handle/uniba/14614 LA - deu ER - TY - CPAPER TI - Compressed Domain Image Retrieval Using JPEG2000 and Gaussian Mixture Models AU - Teynor, Alexandra AU - Müller, Wolfgang AU - Kowarschick, Wolfgang PY - 2005 T2 - Visual information and information systems : 8th international conference, VISUAL 2005, Amsterdam, The Netherlands, July 5, 2005 ; revised selected papers SP - 132 - 142 CY - Berlin UR - https://fis.uni-bamberg.de/handle/uniba/14567 LA - eng ER - TY - JOUR TI - Inhaltsbasierte Suche in P2P-Systemen AU - Eisenhardt, Martin AU - Henrich, Andreas AU - Müller, Wolfgang PY - 2005 JO - Datenbank-Spektrum : Zeitschrift für Datenbanktechnologie und Information Retrieval ; Organ der Fachgruppe Datenbanken der Gesellschaft für Informatik e.V. VL - 5 IS - 12 = Themenschwerpunkt: Web-Suche und Web-Caching SP - 5-15 PB - Springer CY - Berlin ; Heidelberg SN - 1618-2162 UR - https://fis.uni-bamberg.de/handle/uniba/14569 LA - deu AB - Pcer-to-Peer-Netze ( P2P-Netze) erfreuen sich zunehmender Beliebtheit hei der verteilten und administrationsarmen Verwaltung größerer Dokumentenkollektionen. Ein wichtiges Problem ist dabei die Suche und insbesondere die inhaltsbasierte Suche nach Dokumenten in P2P-Netzen. Der vorliegende Beitrag gibt einen Überblick über entsprechende Ansätze und zeigt interessante Forschungsaspekte auf. KW - P2P-Systeme ER - TY - CPAPER TI - Scalable summary based retrieval in P2P networks AU - Müller, Wolfgang AU - Eisenhardt, Martin AU - Henrich, Andreas PY - 2005 T2 - CIKM '05 : Proceedings of the 14th ACM international conference on Information and knowledge management SP - 586-593 PB - ACM CY - New York SN - 978-1-59593-140-5 DO - 10.1145/1099554.1099706 UR - https://fis.uni-bamberg.de/handle/uniba/14571 LA - deu AB - Much of the present P2P-IR literature is focused on distributed indexing structures. Within this paper, we present an approach based on the replication of peer data summaries via rumor spreading and multicast in a structured overlay.We will describe Rumorama, a P2P framework for similar-ity queries inspired by GlOSS and CORI and their P2P-adaptation, PlanetP. Rumorama achieves a hierarchization of PlanetP-like summary-based P2P-IR networks. In a Rumorama network, each peer views the network as a small PlanetP network with connections to peers that see other small PlanetP networks. One important aspect is that each peer can choose the size of the PlanetP network it wants to see according to its local processing power and bandwidth. Even in this adaptive environment, Rumorama manages to process a query such that the summary of each peer is considered exactly once in a network without churn. However, the actual number of peers to be contacted for a query is a small fraction of the total number of peers in the network.Within this article, we present the Rumorama base protocol, as well as experiments demonstrating the scalability and viability of the approach under churn. KW - P2P KW - Similarity Search KW - Distributed IR ER - TY - CPAPER TI - Reducing I/O Cost of Similarity Queries by Processing Several at a Time AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2004 T2 - Proceedings MDDE '04, 4th International Workshop on Multimedia Data and Document Engineering, Washington, DC, USA, July 2nd 2004. IEEE Computer Society 2004 UR - https://fis.uni-bamberg.de/handle/uniba/14663 LA - eng ER - TY - CPAPER TI - Efficient content-based P2P image retrieval using peer content descriptions AU - Müller, Wolfgang AU - Eisenhardt, Martin AU - Henrich, Andreas PY - 2004 T2 - Internet Imaging V : Proceedings of Electronic Imaging, Science and Technology 2004 SP - 57-68 PB - SPIE CY - Bellingham, Wash., USA T3 - SPIE Proceedings SN - 0-8194-5207-6 DO - 10.1117/12.531184 UR - https://fis.uni-bamberg.de/handle/uniba/14609 LA - eng AB - Peer-to-peer (P2P) networks are overlay networks that connect independent computers (also called nodes or peers). In contrast to client/server solutions, all nodes offer and request services from other peers in a P2P network. P2P networks are very attractive in that they harness the computing power of many common desktop machines and necessitate little administrative overhead. While the resulting computing power is impressive, efficiently looking up data still is the major challenge in P2P networks. Current work comprises fast lookup of one-dimensional values (Distributed Hash Tables, DHT) and retrieval of texts using few keywords. However, the lookup of multimedia data in P2P networks is still attacked by very few groups. In this paper, we present experiments with efficient Content Based Image Retrieval in a P2P environment, thus a P2P-CBIR system. The challenge in such systems is to limit the number of messages sent, and to maximize the usefulness of each peer contacted in the query process. We achieve this by distributing peer data summaries over the network. Obviously, the data summaries have to be compact in order to limit the communication overhead. We propose an CBIR scheme based on a compact peer data summary. This peer data summary relies on cluster frequencies. To obtain the compact representation of a peer's collection, a. global clustering of the data is efficiently calculated in a distributed manner. After that, each peer publishes how many of its images fall into each cluster. These cluster frequencies are then used by the querying peer to contact only those peers that have the largest number of images present in one cluster given by the query. In our paper we further detail the various challenges that have to be met by the designers of such a P2P- CBIR, and we present experiments with varying degree of data replication (duplicates of images) , as well as quality of clustering within the network. KW - P2P ER - TY - CPAPER TI - Faster exact histogram intersection on large data collections using inverted VA-files AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2004 T2 - Image and video retrieval : third International Conference, CIVR 2004, Dublin, Ireland, July 21 - 23, 2004 ; proceedings SP - 455-463 PB - Springer CY - Berlin T3 - Lecture notes in computer science SN - 978-3-540-22539-3 DO - 10.1007/978-3-540-27814-6_54 UR - https://fis.uni-bamberg.de/handle/uniba/12774 LA - eng AB - Most indexing structures for high-dimensional vectors used in multimedia retrieval today rely on determining the importance of each vector component at indexing time in order to create the index. However for Histogram Intersection and other important distance measures this is not possible because the importance of vector components depends on the query. We present an indexing structure inspired by VA-file and Inverted file that does not need to determine the importance at indexing time in order to perform well. Instead, our structure adapts to the importance of vector components at query processing time. Success of this approach is demonstrated in experiments on feature data extracted from a large image collection. KW - histogram intersection KW - large data collections ER - TY - CPAPER TI - Spektrale Bloom-Filter für Peer-to-Peer Information Retrieval AU - Eisenhardt, Martin AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2004 T2 - Informatik 2004 : Informatik verbindet ; Beiträge der 34. Jahrestagung der Gesellschaft für Informatik e.V. (GI) ; Ulm, 20. - 24. September 2004 SP - 44 - 48 PB - Gesellschaft für Informatik e.V. CY - Bonn SN - 3-88579-380-6 UR - https://fis.uni-bamberg.de/handle/uniba/12728 LA - deu AB - Spektrale Bloomfilter können als Zusammenfassung von Peer-Daten in verteiltem P2P-Retrieval verwendet werden. Dies ermöglicht eine messbar bessere Auswahl derjenigen Peers, an die eine Anfrage gestellt wird, und somit auch eine höhere Effizienz der Anfragebearbeitung. Experimentelle Daten belegen diese Verbesserung des Retrievals in P2P-Netzen. KW - Spektrale Bloom-Filter ER - TY - CPAPER TI - Fast Retrieval of High-Dimensional Feature Vectors in P2P Networks Using Compact Peer Data Summaries AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2003 T2 - MIR '03 : Proceedings of the 5th ACM SIGMM international workshop on Multimedia information retrieval SP - 79–86 PB - ACM CY - New York SN - 978-1-58113-778-1 DO - 10.1145/973264.973278 UR - https://fis.uni-bamberg.de/handle/uniba/14659 LA - deu AB - The retrieval facilities of most Peer-to-Peer (P2P) systems are limited to queries based on a unique identifier or a small set of keywords. The techniques used for this purpose are hardly applicable for content-based image retrieval (CBIR) in a P2P network. Furthermore, we will argue that the curse of dimensionality and the high communication overhead prevent the adaptation of multidimensional search trees or fast sequential scan techniques for P2P CBIR. In the present paper we will propose two compact data representations which can be distributed in a P2P network and used as the basis for a source selection. This allows to communicate only with a small fraction of all peers during query processing without deteriorating the result quality significantly. We will also present experimental results confirming our approach. KW - Algorithms KW - Performance ER - TY - CHAP TI - Was leisten automatische inhaltsbasierte Bildsuchsysteme? AU - Marchand-Maillet, Stéphane AU - Lasri, Nadia AU - Müller, Henning AU - Müller, Wolfgang AU - Pun, Thierry PY - 2003 T2 - Suchbilder : visuelle Kultur zwischen Algorithmen und Archiven; im Verbund mit dem Symposion "Suchbilder. Schritte zu einem Bildarchiv Filmischer Topoi" entstanden (6. bis 8. Februar 2001 in den Kunst- Werken Berlin) SP - 86-98 PB - Kulturverlag Kadmos CY - Berlin SN - 978-3-931659-35-6 UR - https://fis.uni-bamberg.de/handle/uniba/14638 LA - deu ER - TY - CPAPER TI - Comparing the performance of two CBIRS indexing schemes AU - Müller, Wolfgang AU - Robbert, Günter AU - Henrich, Andreas PY - 2003 T2 - Internet Imaging IV : Proceedings of Electronic Imaging, Science and Technology 2003 SP - 9-20 PB - SPIE CY - Bellingham, Wash., USA T3 - SPIE Proceedings SN - 0-8194-4818-4 DO - 10.1117/12.473372 UR - https://fis.uni-bamberg.de/handle/uniba/14612 LA - eng AB - Content based image retrieval (CBIR) as it is known today has to deal with a number of challenges. Quickly summarized, the main challenges are firstly, to bridge the semantic gap between high-level concepts and low-level features using feedback, secondly to provide performance under adverse conditions. High-dimensional spaces, as well as a demanding machine learning task make the right way of indexing an important issue. When indexing multimedia data, most groups opt for extraction of high-dimensional feature vectors from the data, followed by dimensionality reduction like PCA (Principal Components Analysis) or LSI (Latent Semantic Indexing). The resulting vectors are indexed using spatial indexing structures such as kd-trees or R-trees, for example. Other projects, such as MARS and Viper propose the adaptation of text indexing techniques, notably the inverted file. Here, the Viper system is the most direct adaptation of text retrieval techniques to quantized vectors. However, while the Viper query engine provides decent performance together with impressive user-feedback behavior, as well as the possibility for easy integration of long-term learning algorithms, and support for potentially infinite feature vectors, there has been no comparison of vector-based methods and inverted-file-based methods under similar conditions. In this publication, we compare a CBIR query engine that uses inverted files (Bothrops, a rewrite of the Viper query engine based on a relational database), and a CBIR query engine based on LSD (Local Split Decision) trees for spatial indexing using the same feature sets. The Benchathlon initiative works on providing a set of images and ground truth for simulating image queries by example and corresponding user feedback. ·when performing the Benchathlon benchmark on a CBIR system (the System Under Test, SUT), a benchmarking harness connects over internet to the SUT, performing a number of queries using an agreed-upon protocol, the multimedia retrieval markup language (MRML). Using this bench-mark one can measure the quality of retrieval, as well as the overall (speed) performance of the benchmarked system. Our Benchmarks will draw on the Benchathlon's work for documenting the retrieval performance of both inverted file-based and LSD tree based techniques. However, in addition to these results, we will present statistics, that can be obtained only inside the system under test. These statistics will include the number of complex mathematical operations, as well as the amount of data that has to be read from disk during operation of a query. KW - CBIRS indexing schemes ER - TY - CPAPER TI - Privacy of Ideas in P2P Networks AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2003 T2 - PET2003 Dresden, Germany ; Workshop on Privacy Enhancing Technologies, Hotel Elbflorenz Dresden, Germany, 26 - 28 March 2003 / Program / Wednesday, March 26 / Rump session / works in progress (part one) UR - https://fis.uni-bamberg.de/handle/uniba/14613 LA - deu ER - TY - CPAPER TI - A framework for benchmarking in CBIR AU - Müller, Henning AU - Müller, Wolfgang AU - Marchand-Maillet, Stéphane AU - Squire, David McG. AU - Pun, Thierry PY - 2003 VL - 21 IS - 1 = Special Issue: Best Papers of the ACM Multimedia 2001 Workshop on Multimedia Information Retrieval SP - 55 - 73 SN - 1380-7501 UR - https://fis.uni-bamberg.de/handle/uniba/14608 LA - eng ER - TY - BOOK TI - Classifying Documents by Distributed P2P Clustering AU - Eisenhardt, Martin AU - Müller, Wolfgang AU - Henrich, Andreas PY - 2003 CY - Bamberg UR - https://fis.uni-bamberg.de/handle/uniba/14605 LA - eng ER - TY - JOUR TI - Bildersuchbaukasten - Das GNU Image Finding Tool via Plug-ins erweitern AU - Müller, Wolfgang PY - 2002 JO - C't : Magazin für Computer-Technik IS - 17 SP - 190 SN - 0724-8679 UR - https://fis.uni-bamberg.de/handle/uniba/14604 LA - deu ER - TY - CPAPER TI - A Web-Based Evaluation System for Content-Based Image Retrieval AU - Müller, Henning AU - Müller, Wolfgang AU - Marchand-Maillet, Stéphane AU - Squire, David McG. AU - Pun, Thierry PY - 2001 T2 - Proceedings of ACM Multimedia 2001 workshops CY - New York, NY, UR - https://fis.uni-bamberg.de/handle/uniba/14630 LA - eng ER - TY - CHAP TI - Design and Evaluation of a Content-based Image Retrieval System AU - Squire, David McG. AU - Müller, Henning AU - Müller, Wolfgang AU - Marchand-Maillet, Stéphane AU - Pun, Thierry PY - 2001 T2 - Design and management of multimedia information systems : opportunities and challenges CY - Hershey u.a. SN - 9781930708006 UR - https://fis.uni-bamberg.de/handle/uniba/14634 LA - eng ER - TY - CPAPER TI - Automated benchmarking in content-based image retrieval AU - Müller, Henning AU - Müller, Wolfgang AU - Marchand-Maillet, Stéphane AU - McG.Squire, David AU - Pun, Thierry PY - 2001 T2 - 2001 IEEE International Conference on Multimedia and Expo, ICME 2001 : 22 - 25 August 2001, Tokyo, Japan ; Theme: Recent advances in multimedia world CY - Piscataway, NJ UR - https://fis.uni-bamberg.de/handle/uniba/14590 LA - deu ER - TY - BOOK TI - Automated benchmarking in content-based image retrieval AU - Müller, Henning AU - Müller, Wolfgang AU - Marchand-Maillet, Stéphane AU - McG.Squire, David AU - Pun, Thierry PY - 2001 CY - Geneve T3 - Technical report vision SN - 0-7695-1198-8 UR - https://fis.uni-bamberg.de/handle/uniba/14589 LA - deu ER - TY - JOUR TI - Evaluating Image Browsers Using Structured Annotation AU - Müller, Wolfgang AU - Marchand-Maillet, Stéphane AU - Müller, Henning AU - McG.Squire, David AU - Pun, Thierry PY - 2001 JO - Journal of the American Society for Information Science and Technology : JASIST VL - 52 IS - 11 SP - 961 - 968 SN - 1532-2882 UR - https://fis.uni-bamberg.de/handle/uniba/14594 LA - eng ER - TY - JOUR TI - Performance evaluation in content-based image retrieval: overview and proposals AU - Müller, Henning AU - Müller, Wolfgang AU - Squire, David McG. AU - Marchand-Maillet, Stéphane AU - Pun, Thierry PY - 2001 JO - Pattern recognition letters : an official publ. of the International Association for Pattern Recognition VL - 22 IS - 5 SP - 593 - 601 SN - 0167-8655 UR - https://fis.uni-bamberg.de/handle/uniba/14595 LA - eng ER -