Hierarchy Identification for Automatically Generating Table-of-Contents

Proceedings of Recent Advances in Natural Language Processing, pages 252–260, Hissar, Bulgaria, 7-13 September 2013. Hierarchy Identification for Automatically Generating Table-of-Contents Nicolai Erbsα αUbiquitous Knowledge Processing Lab Department of Computer Science, Technische Universit¨at Darmstadt Iryna Gurevychαβ βInformation Center for Education German Institute for Educational Research and Educational Information Torsten Zeschγ γLanguage Technology University of Duisburg-Essen Abstract A table-of-contents (TOC) provides a quick reference to a document’s content and structure. We present the first study on identifying the hierarchical structure for automatically generating a TOC us- ing only textual features instead of struc- tural hints e.g. from HTML-tags. We cre- ate two new datasets to evaluate our ap- proaches for hierarchy identification. We find that our algorithm performs on a level that is sufficient for a fully automated sys- tem. For documents without given seg- ment titles, we extend our work by auto- matically generating segment titles. We make the datasets and our experimen- tal framework publicly available in order to foster future research in TOC genera- tion. 1 Introduction A table-of-contents (TOC) provides an easy way to gain an overview about a document as a TOC presents the document’s content and structure. At the same time, a TOC captures the relative impor- tance of document topics by arranging the topic titles in a hierarchical manner. Thus, TOCs might be used as a short document summary that pro- vides more information about search results in a search engine. Figure 1 provides a sketch of such a search interface. Instead of a thumbnail of the document like most search engines, or a cluster- ing of search results (Carpineto et al., 2009), we propose to use an automatically extracted TOC. The task of automatically generating a table-of- contents can be tackled with the subtasks docu- ment segmentation, segment title generation, and hierarchy identification. The first step splits the document into topical parts, the second step gen- erates an informative title for each segment, and Search sorting algorithms go Sorting algorithm - Wikipedia, the free encyclopedia en.wikipedia.org/wiki/Sorting_algorith m In computer science, a sorting algorithm is an algorithm that puts elements of a list in a certain order. The most-used orders are numerical order and ... Quicksort - Bubble sort - Merge sort - Shellsort Sorting Algorithm Animations www.sorting-algorithms.com/ Animation, code, analysis, and discussion of 8 sorting algorithms on 4 initial conditions. Quick Sort - Insertion Sort - Quick Sort (3 Way Partition) - Shell Sort Sorting Algorithms cs.smith.edu/~thiebaut/java/sort/demo. html... 1 Classification 1.1 Stability 2 Comparison of algorithms 3 Summaries of popular sorting algorithms 3.1 Bubble sort 3.2 Selection sort 3.3 Insertion sort 3.4 Shell sort 3.5 Comb sort 3.6 Merge sort 3.7 Heapsort 3.8 Quicksort 3.9 Counting sort 3.10 Bucket sort 3.11 Radix sort 3.12 Distribution sort 3.13 Timsort 4 Memory usage patterns and index sorting 5 Inefficient/humorous sorts 6 See also 7 References 8 External links Figure 1: Search user interface showing a TOC along with the search results. the third step decides whether a segment is on a higher, equal, or lower …

Content truncated to 500 words