[ { "id": "mazurek-2018-vac", "type_id": "masterthesis", "tu_id": 282668, "repositum_id": null, "title": "Visual Active Learning for News Stream Classification", "date": "2019-10-10", "abstract": "In many domains, the sheer quantity of text documents that have to be parsed increases\ndaily. To keep up with this continuous text stream, a considerable amount of time has to be invested. We developed a classification interface for text streams that learns user-specific topics from the user’s labeling process and partitions the incoming data into these topics.\nCurrent approaches that try to derive content categorization from a vast number of unstructured text documents use pre-trained learning models to perform text classification.\nThese models assign predefined categories to the text according to its content. Depending on the use case, a user’s interests might not coincide with the given categories. The model cannot adapt to changing terminology that was not present during training. Besides these factors, users often do not trust pre-trained models as they are a black box for them.\nTo solve this problem, our application lets users define a classification problem and\ntrain a learning model through interaction with a Star Coordinates visualization. The\napproach that makes this interaction efficient is a variant of active learning. This active learning variant states that a learning model can achieve greater accuracy with fewer labeled training instances, if a user provides data purposefully from which it learns. We adapted this strategy for text stream classification by visualizing the topic affiliation probabilities of the learning model and providing novel interaction tools to enhance the model’s performance iteratively. By simulating different selection strategies common in active learning, we found that our visual selection strategies correspond closely to the classic active learning selection\nstrategies. Further, users performed on par with the best simulated selection strategies in the results from our preliminary user study. Our evaluation concludes that there are benefits from incorporating information visualization into the active learning process.", "authors_et_al": false, "substitute": null, "main_image": { "description": null, "filetitle": "image", "main_file": true, "use_in_gallery": true, "access": "public", "image_width": 668, "image_height": 638, "name": "mazurek-2018-vac-image.png", "type": "image/png", "size": 56667, "path": "Publication:mazurek-2018-vac", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/mazurek-2018-vac/mazurek-2018-vac-image.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/mazurek-2018-vac/mazurek-2018-vac-image:thumb{{size}}.png" }, "sync_repositum_override": null, "repositum_presentation_id": null, "authors": [ 1364 ], "date_end": "2019-10-10", "date_start": "2018-03", "matrikelnr": "0123456", "open_access": "yes", "supervisor": [ 1110, 166 ], "research_areas": [ "InfoVis" ], "keywords": [], "weblinks": [], "files": [ { "description": null, "filetitle": "image", "main_file": true, "use_in_gallery": true, "access": "public", "image_width": 668, "image_height": 638, "name": "mazurek-2018-vac-image.png", "type": "image/png", "size": 56667, "path": "Publication:mazurek-2018-vac", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/mazurek-2018-vac/mazurek-2018-vac-image.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/mazurek-2018-vac/mazurek-2018-vac-image:thumb{{size}}.png" }, { "description": null, "filetitle": "Master thesis", "main_file": true, "use_in_gallery": true, "access": "public", "name": "mazurek-2018-vac-Master thesis.pdf", "type": "application/pdf", "size": 5490435, "path": "Publication:mazurek-2018-vac", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/mazurek-2018-vac/mazurek-2018-vac-Master thesis.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/mazurek-2018-vac/mazurek-2018-vac-Master thesis:thumb{{size}}.png" }, { "description": null, "filetitle": "Poster", "main_file": true, "use_in_gallery": true, "access": "public", "name": "mazurek-2018-vac-Poster.pdf", "type": "application/pdf", "size": 1520895, "path": "Publication:mazurek-2018-vac", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/mazurek-2018-vac/mazurek-2018-vac-Poster.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/mazurek-2018-vac/mazurek-2018-vac-Poster:thumb{{size}}.png" } ], "projects_workgroups": [ "deskollage" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2019/mazurek-2018-vac/", "__class": "Publication" }, { "id": "mazurek-2018-veq", "type_id": "journalpaper", "tu_id": null, "repositum_id": null, "title": "Visualizing Expanded Query Results", "date": "2018-06", "abstract": "When performing queries in web search engines, users often face difficulties choosing appropriate query terms. Search engines\ntherefore usually suggest a list of expanded versions of the user query to disambiguate it or to resolve potential term mismatches.\nHowever, it has been shown that users find it difficult to choose an expanded query from such a list. In this paper, we describe\nthe adoption of set-based text visualization techniques to visualize how query expansions enrich the result space of a given\nuser query and how the result sets relate to each other. Our system uses a linguistic approach to expand queries and topic\nmodeling to extract the most informative terms from the results of these queries. In a user study, we compare a common text list\nof query expansion suggestions to three set-based text visualization techniques adopted for visualizing expanded query results\n– namely, Compact Euler Diagrams, Parallel Tag Clouds, and a List View – to resolve ambiguous queries using interactive\nquery expansion. Our results show that text visualization techniques do not increase retrieval efficiency, precision, or recall.\nOverall, users rate Parallel Tag Clouds visualizing key terms of the expanded query space lowest. Based on the results, we derive\nrecommendations for visualizations of query expansion results, text visualization techniques in general, and discuss alternative\nuse cases of set-based text visualization techniques in the context of web search.", "authors_et_al": false, "substitute": null, "main_image": { "description": "Density-based compact Euler Diagram", "filetitle": "teaser", "main_file": false, "use_in_gallery": true, "access": "public", "image_width": 919, "image_height": 656, "name": "mazurek-2018-veq-teaser.png", "type": "image/png", "size": 34882, "path": "Publication:mazurek-2018-veq", "url": "https://www.cg.tuwien.ac.at/research/publications/2018/mazurek-2018-veq/mazurek-2018-veq-teaser.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2018/mazurek-2018-veq/mazurek-2018-veq-teaser:thumb{{size}}.png" }, "sync_repositum_override": null, "repositum_presentation_id": null, "authors": [ 1364, 1110 ], "cfp": { "name": "EuroVis 2018 CfP.pdf", "type": "application/pdf", "error": "0", "size": "55844", "orig_name": "EuroVis 2018 CfP.pdf", "ext": "pdf" }, "date_from": "2018-06-04", "date_to": "2018-06-08", "event": "EG / VGTC Conference on Visualization (EuroVis)", "journal": "Computer Graphics Forum", "lecturer": [ 1364 ], "location": "Brno, Czech Republic", "pages_from": "87", "pages_to": "98", "research_areas": [ "InfoVis", "NetVis" ], "keywords": [ "Information visualization", "search interfaces", "empirical studies in visualization" ], "weblinks": [], "files": [ { "description": null, "filetitle": "paper", "main_file": true, "use_in_gallery": false, "access": "public", "name": "mazurek-2018-veq-paper.pdf", "type": "application/pdf", "size": 1891501, "path": "Publication:mazurek-2018-veq", "url": "https://www.cg.tuwien.ac.at/research/publications/2018/mazurek-2018-veq/mazurek-2018-veq-paper.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2018/mazurek-2018-veq/mazurek-2018-veq-paper:thumb{{size}}.png" }, { "description": "Density-based compact Euler Diagram", "filetitle": "teaser", "main_file": false, "use_in_gallery": true, "access": "public", "image_width": 919, "image_height": 656, "name": "mazurek-2018-veq-teaser.png", "type": "image/png", "size": 34882, "path": "Publication:mazurek-2018-veq", "url": "https://www.cg.tuwien.ac.at/research/publications/2018/mazurek-2018-veq/mazurek-2018-veq-teaser.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2018/mazurek-2018-veq/mazurek-2018-veq-teaser:thumb{{size}}.png" }, { "description": null, "filetitle": "video", "main_file": true, "use_in_gallery": false, "access": "public", "name": "mazurek-2018-veq-video.mp4", "type": "video/mp4", "size": 30231176, "path": "Publication:mazurek-2018-veq", "url": "https://www.cg.tuwien.ac.at/research/publications/2018/mazurek-2018-veq/mazurek-2018-veq-video.mp4", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2018/mazurek-2018-veq/mazurek-2018-veq-video:thumb{{size}}.png", "video_mp4": "https://www.cg.tuwien.ac.at/research/publications/2018/mazurek-2018-veq/mazurek-2018-veq-video:video.mp4" } ], "projects_workgroups": [ "deskollage" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2018/mazurek-2018-veq/", "__class": "Publication" }, { "id": "mazurek-2017-sio", "type_id": "bachelorthesis", "tu_id": null, "repositum_id": null, "title": "Stream I/O - An Interactive Visualization of Publication Data", "date": "2017", "abstract": "The publication database of the Institute of Computer Graphics and Algorithms can\ncurrently be queried by a simple UI which returns a list. Stream I/O, the application\nof this thesis, extends the interface to improve it in terms of overview, exploration and\nanalysis support. To cope with these shortcommings a visualization is added to the user\ninterface. As the publication database includes a lot of additional data attributes, a\nselection of attributes is used for the visualization to give further insight. By using the\nStreamgraph [BW08] visualization, the variations over time within attributes like authors,\npublication type and research areas are made visible. The focus of this visualization lies\nin showing individual attribute values while also conveying the sum. This relationship\nis depicted in a timeline, which allows a user to explore the past and current work of\nthe institute as well as to find relationships and trends in the publications. As the\nvisualization uses a timeline encoding, the directed flow from left to right is interpreted\nas the movement through time. It shows the evolution of different attributes, while the\noccurrence of a topic at a specific time is coded with the width of the layer at a specific\npoint. Searching the database is enriched through multiple viewpoints which give the user\ninsight how attributes relate in the underlying data and how the data is changing through\nhis/her manipulation. Selections of colored layers within the graph can represent bigger\ntrends and give insight into the data as a whole. The Stream I/O application invites\nusers to interactively explore the publication database, while simultaneously gaining new\ninsight through the visualization.", "authors_et_al": false, "substitute": null, "main_image": { "description": null, "filetitle": "", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 1293, "image_height": 562, "name": "mazurek-2017-sio-.png", "type": "image/png", "size": 403642, "path": "Publication:mazurek-2017-sio", "url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-sio/mazurek-2017-sio-.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-sio/mazurek-2017-sio-:thumb{{size}}.png" }, "sync_repositum_override": null, "repositum_presentation_id": null, "authors": [ 1364 ], "date_end": "2017-03", "date_start": "2016-05", "duration": "4 months", "matrikelnr": "0123456", "supervisor": [ 1110 ], "research_areas": [ "InfoVis" ], "keywords": [], "weblinks": [], "files": [ { "description": null, "filetitle": "", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 1293, "image_height": 562, "name": "mazurek-2017-sio-.png", "type": "image/png", "size": 403642, "path": "Publication:mazurek-2017-sio", "url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-sio/mazurek-2017-sio-.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-sio/mazurek-2017-sio-:thumb{{size}}.png" }, { "description": null, "filetitle": "thesis", "main_file": false, "use_in_gallery": false, "access": "public", "name": "mazurek-2017-sio-thesis.pdf", "type": "application/pdf", "size": 4004622, "path": "Publication:mazurek-2017-sio", "url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-sio/mazurek-2017-sio-thesis.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-sio/mazurek-2017-sio-thesis:thumb{{size}}.png" } ], "projects_workgroups": [ "deskollage" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-sio/", "__class": "Publication" }, { "id": "mazurek-2017-vows", "type_id": "studentproject", "tu_id": null, "repositum_id": null, "title": "Visualization of Thesaurus-Based Web Search", "date": "2017", "abstract": "The general functions of current web search engines are well established. A box is\nprovided in which to type the queries and the engine returns a result list which users can\nevaluate. The autocomplete suggestions assist users in defining their problems, however\nthere is a lack of support for an iterative manual refinement of the query. This additional\naid can be beneficial when users not know the exact terms to describe the concept they\nare looking for. Therefore, the goal of this project is to show searchers how a slight\nvariation of the query changes the results. With this information, they then can perform\na targeted refinement of the query to access useful information sources. To achieve this\ngoal, each part of the searcher’s query is varied with a thesaurus that provides synonyms\nfor the individual query terms. While performing the user’s original query in a normal\nfashion, variations of this query are conducted in the background. To provide a concise\nvisual summary of the query results, text mining techniques are performed on all gathered\nresults to retrieve the most important key terms for each query variation. This procedure\nresults in a visual overview of what the searcher’s query finds together with what could\nbe found with a slight variation of the query. Additional queries should make users aware\nthat alternative queries may be more appropriate when their original query is poorly\nformulated. In conjunction with some interaction tools, the goal is to reduce the burden\nof refining search queries and therefore making searching the web less complex.", "authors_et_al": false, "substitute": null, "main_image": { "description": null, "filetitle": "teaser", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 814, "image_height": 539, "name": "mazurek-2017-vows-teaser.PNG", "type": "image/png", "size": 405860, "path": "Publication:mazurek-2017-vows", "url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-vows/mazurek-2017-vows-teaser.PNG", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-vows/mazurek-2017-vows-teaser:thumb{{size}}.png" }, "sync_repositum_override": null, "repositum_presentation_id": null, "authors": [ 1364 ], "date_end": "2017-10", "date_start": "2017-08", "matrikelnr": "1126483", "supervisor": [ 1110 ], "research_areas": [ "InfoVis" ], "keywords": [], "weblinks": [], "files": [ { "description": null, "filetitle": "report", "main_file": false, "use_in_gallery": false, "access": "public", "name": "mazurek-2017-vows-report.pdf", "type": "application/pdf", "size": 540973, "path": "Publication:mazurek-2017-vows", "url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-vows/mazurek-2017-vows-report.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-vows/mazurek-2017-vows-report:thumb{{size}}.png" }, { "description": null, "filetitle": "teaser", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 814, "image_height": 539, "name": "mazurek-2017-vows-teaser.PNG", "type": "image/png", "size": 405860, "path": "Publication:mazurek-2017-vows", "url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-vows/mazurek-2017-vows-teaser.PNG", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-vows/mazurek-2017-vows-teaser:thumb{{size}}.png" } ], "projects_workgroups": [ "deskollage" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2017/mazurek-2017-vows/", "__class": "Publication" } ]