Natural Language Generation Pipeline

NLG Pipeline
 1. Data ingestion

  • data scraping
  • web scraping
 2. Data analytics
 3. Data interpretation
 4. Macro-planning

  • document orchestration
  • document planning
 5. Micro-planning
 6. Realization

  • surface realization
 7. Publishing

  • surface text

 

Notes:

A natural language generation pipeline is a series of steps or stages that are followed in order to generate human-like text from data or other input. The specific components of a natural language generation pipeline can vary depending on the specific application and the desired output, but in general, a typical pipeline might include the following steps:

  1. Data preparation: In this stage, the input data is cleaned, organized, and prepared for use by the natural language generation system. This might involve tasks such as removing unnecessary or irrelevant data, formatting the data in a specific way, or extracting specific pieces of information that will be used to generate the output text.
  2. Content determination: In this stage, the system determines what information from the input data should be included in the output text. This might involve tasks such as identifying the most important or relevant pieces of information, selecting the appropriate tone or style for the output text, or determining the overall structure or organization of the text.
  3. Sentence generation: In this stage, the system generates individual sentences or phrases based on the information determined in the previous stage. This might involve tasks such as selecting the appropriate words and phrases to use, combining them in a grammatically correct way, and applying the appropriate punctuation.
  4. Text aggregation: In this stage, the system combines the individual sentences or phrases generated in the previous stage into a coherent and cohesive piece of text. This might involve tasks such as ensuring that the text flows smoothly from one sentence to the next, ensuring that the text has the desired tone or style, and adding any necessary transitions or connectors.
  5. Text refinement: In this final stage, the system performs any necessary final adjustments to the generated text. This might involve tasks such as proofreading the text for grammar and spelling errors, ensuring that the text adheres to any specified style guidelines, or making any necessary changes to improve the overall quality or clarity of the text.

Overall, a natural language generation pipeline is a complex and multi-faceted process that involves a wide range of techniques and technologies to generate human-like text from data or other input. By following these steps, a natural language generation system can produce high-quality, human-like text that is tailored to the specific needs and requirements of the application.

  • Data-to-text
  • Natural language generator
  • NLG pipeline
  • NLG pipelines
  • International Natural Language Generation (INLG) conference
  • Virtual storyteller

Wikipedia:

See also:

KPML (Komet-Penman Multi-Lingual) | Natural Language Generation Engines | Realizers In Natural Language Processing | Sentence Generation Module | Sentence Planner | SimpleNLG Realization Engine

Spartiqulation: Verbalizing sparql queries B Ell, D Vrandecic, E Simperl – The Semantic Web: ESWC 2012 Satellite …, 2012 – Springer … Message types are classified due to their role within the verbalization. The document plan is presented which orders the messages according to their classes and is the output of the Document Planner – the first component in our NLG pipeline. … Cited by 19 Related articles All 9 versions

Generating different story tellings from semantic representations of narrative E Rishes, SM Lukin, DK Elson, MA Walker – Interactive Storytelling, 2013 – Springer … graph. This represen- tation is robust and linguistically grounded which makes it a good candidate for a content representation in an nlg pipeline. Page 3. 194 E. Rishes et al. … personage. We integrated these tools in a standard nlg pipeline: … Cited by 15 Related articles All 7 versions

Towards generating text from discourse representation structures V Basile, J Bos – Proceedings of the 13th European Workshop on …, 2011 – dl.acm.org … of the DRSs in this corpus is, and what phe- nomena are covered (Section 3). We also illustrate what challenges it poses upon micro planning and surface realisation (Section 4). Finally, in Section 5, we discuss how generating from DRSs relates to the traditional NLG pipeline. … Cited by 10 Related articles All 13 versions

Generating and interpreting referring expressions as belief state planning and plan recognition DA Smith, H Lieberman – ENLG 2013, 2013 – aclweb.org … targets from the dis- tractors. This content is then passed to the next module in an NLG pipeline (cf (Reiter, 1994)) to ultimately become a noun phrase embedded in a larger construct. These “pipeline” architectures prevent infor … Cited by 5 Related articles All 12 versions

Exploiting ontology lexica for generating natural language texts from RDF data P Cimiano, J Lüker, D Nagel, C Unger – 2013 – pub.uni-bielefeld.de … While it follows a classical NLG pipeline, it diverges from most cur- rent NLG systems in that it exploits an ontology lexicon in order to capture context-specific lexicalisations of ontol- ogy concepts, and combines the use of such a lexicon with the choice of lexical items and … Cited by 8 Related articles All 11 versions

A natural language ontology-driven query interface E Franconi, P Guagliardo, S Tessaris… – … on Terminology and …, 2011 – perso.limsi.fr … While in the classic NLG pipeline the infor- mation to be conveyed in the text and its order is determined by the document planning module, in the Query Tool it is the user who decides both the information to be displayed and its arrangement. … Cited by 2 Related articles All 8 versions

Joint morphological generation and syntactic linearization L Song, Y Zhang, K Song, Q Liu – Proceedings of the Twenty- …, 2014 – researchgate.net … While most NLG pipelines separate morphological generation and syn- tactic linearization, the two tasks are closely related. … Sem Synt Morphological String (b) Figure 1: (a) NLG pipeline based on the meaning text theory (Bohnet et al. 2010), (b) pipeline of this paper. … Cited by 4 Related articles All 3 versions

Reading and writing as a creative cycle: the need for a computational model P Gervás, C León – 5th International Conference on Computational …, 2014 – nil.fdi.ucm.es … The content determina- tion stage of a NLG pipeline assumes all possible content to be present, and applies a selection process to establish what will be included in the communication under consideration. … In truth, all of the NLG pipeline could … Cited by 6 Related articles All 5 versions

Using multimodal resources for explanation approaches in technical systems F Nothdurft, W Minker – Proc. of the 8th Conf. on Intl. Language …, 2012 – lrec.elra.info … 414 Page 5. Text generation is done by using a typical natural language generation (NLG) pipeline (Reiter and Dale, 2000). The first step in a NLG-pipeline, namely the content determi- nation is realized by the explanation machine. … Cited by 2 Related articles All 2 versions

Abstractive meeting summarization with entailment and fusion Y Mehdad, G Carenini, FW Tompa, RT NG – Proc. of ENLG, 2013 – anthology.aclweb.org … the source meet- ing sentences. In addition, instead of generat- ing an abstractive sentence via the conventional NLG pipeline (Reiter and Dale, 2000), we exploit a word graph approach. Linking Detection Entailment Identify … Cited by 11 Related articles All 11 versions

Toward Natural Language Generation by Humans JO Ryan, AM Fisher, T Owen-Milner, M Mateas… – Proc. INT, 2015 – users.soe.ucsc.edu … The major benefits of NLG are its alleviation of authorial burden and the capability it gives to a system of generating state-bespoke content, but we believe we can have these benefits without ac- tually employing a full NLG pipeline. … Cited by 1

Combining Referring Expression Generation and Surface Realization: A Corpus-Based Investigation of Architectures. S Zarrieß, J Kuhn – ACL (1), 2013 – aclweb.org … (1) a. *John homed him with an order. b. John ordered him home. In recent data-driven generation research, the focus has somewhat shifted from full data-to-text systems to approaches that isolate well-defined subproblems from the NLG pipeline. … Cited by 2 Related articles All 6 versions

Natural Language Generation and Interfaces to Knowledge Bases C Gardent, E Banik… – Invited Tutorial at …, 2011 – computational-linguistics.com … What is NLG? ? The NLG pipeline NLG and Natural Language Interfaces to Knowledge Bases (Claire Gardent and Laura Perez-Beltrachini) ? Verbalising Knowledge Bases ? Querying Knowledge Bases ? Authoring Knowledge Bases Page 11. Table of Contents … Cited by 1 Related articles All 4 versions

Model driven development approach to natural language generation systems H Macedo – ACM SIGSOFT Software Engineering Notes, 2010 – dl.acm.org … RAGS [3], for instance. Figure 1. Fine-grained NLG pipeline … d) and produces linguistically correct final XML-encoded documents (PSMd). These instances undergo NLG pipeline and are conti- nuously rewritten according to the next representation layer. … Cited by 1 Related articles

Natural language generation for a smart biology textbook E Banik, E Kow, V Chaudhri, N Dinesh… – Proceedings of the …, 2012 – dl.acm.org … Generation Lexicon Grammar: Description of TAG tree templates Concept-to-Word mappings Mapping of KB relations to TAG tree templates Morphological lexicon Verb frames (preposition choice) NLG Pipeline Figure 2: Linguistic resources and the generation pipeline … Cited by 1 Related articles All 12 versions

Gen-meta: Generating metaphors using a combination of ai reasoning and corpus-based modeling of formulaic expressions A Gargett, J Barnden – Technologies and Applications of …, 2013 – ieeexplore.ieee.org … At this point, the missing ECG generation compo- nent presents a gap in our NLG pipeline, to be filled by the DS generation mechanism via a two-stage approach: stage 1 translates the Goal Schema into a Goal Tree and switches to the DS generation mechanism, thereby … Cited by 1 Related articles All 5 versions

Generating Annotated Graphs using the NLG Pipeline Architecture S Mahamood, W Bradshaw, E Reiter, NLG Arria – INLG 2014, 2014 – anthology.aclweb.org Abstract The Arria NLG Engine has been extended to generate annotated graphs: data graphs that contain computer-generated textual annotations to explain phenomena in those graphs. These graphs are generated alongside text-only data summaries. Related articles All 8 versions

Natural Language Generation O Dušek – 2013 – ufal.mff.cuni.cz … Ondrej Dušek Natural Language Generation Page 3. Overview Content Planning Surface Realization Standard (Textbook) NLG Pipeline [Input] + Content/Text Planning (“what to say”) ? Content selection, basic ordering [Text plan] … Related articles All 2 versions

Towards a Truly Statistical Natural Language Generator O Dušek – 2013 – ufal.ms.mff.cuni.cz … Ondrej Dušek Towards a Truly Statistical Natural Language Generator Page 3. . . . . . Introduction Statistics in NLG Prospects Standard NLG Pipeline (Textbook) [Input] ? Content/Text Planning (“what to say”) • Content selection, basic structuring (ordering) [Text plan] … Related articles All 3 versions

MinkApp: generating spatio-temporal summaries for nature conservation volunteers N Tintarev, Y Melero, S Sripada, E Tait… – Proceedings of the …, 2012 – dl.acm.org … We identify challenges at all stages of the clas- sical NLG pipeline. 1 Introduction We describe preliminary work on summarizing spatio-temporal data, with the aim to generate con- textualized feedback for wildlife management vol- unteers. … Cited by 1 Related articles All 11 versions

An Introduction To Content Determination G Casamayor, C Mellish – nlgsummer.github.io … Decides “what to say” in terms of domain concepts Page 5. Reiter and Dale (2000) NLG pipeline The application … Domain dependent Language dependent Page 6. NLG pipelines in dialogue systems • See slides on Statistical Natural Language Generation by M.White (2010) …

Towards a Truly Statistical Natural Language Generator for Spoken Dialogues O Dušek – mff.cuni.cz … The Standard NLG Pipeline The standard NLG pipeline as presented in a textbook [Reiter and Dale, 2000], which is at least partially reflected in most today’s NLG systems, consists of two main steps: content planning (or text planning) and surface realization. … Related articles

Using Plan Recognition for Interpreting Referring Expressions. DA Smith, H Lieberman – AAAI Workshop: Plan, Activity, and Intent …, 2013 – media.mit.edu … attribute-value matrix or logical form: s =[TYPE circle COLOR green ] s = TYPE(x0,circle) ? COLOR(x0,green) The content s would then be handed to the next step in an NLG pipeline (cf (Mellish et al. 2004)) with the ultimate goal of … Related articles All 6 versions

Thoughtland: Natural Language Descriptions for Machine Learning n-dimensional Error Functions PA Duboue – ENLG 2013, 2013 – aclweb.org … in Java. It contains a straightforward, easy to understand and modify classic NLG pipeline based on well understood technology like McKeown’s (1985) schemata and Gatt and Reiter’s (2009) SimpleNLG project. This pipeline … Related articles All 9 versions

INLG 2014 M Mitchell, K McCoy, D McDonald, A Cahill – 2014 – anthology.aclweb.org … 118 Generating Annotated Graphs using the NLG Pipeline Architecture Saad Mahamood, William Bradshaw and Ehud Reiter….. 123 Generating Valence Shifted Turkish Sentences Seniz Demir….. … All 7 versions

Towards a programmable instrumented generator C Mellish – Proceedings of the 6th International Natural Language …, 2010 – dl.acm.org … particular stylistic guide- lines. Often conformance can only be tested at the end of the NLG pipeline, when a whole number of complex strategic and tactical decisions have been made, resulting in a complete text. A number of … Related articles All 10 versions

Ontology Authoring Inspired By Dialogue A Parvizi, Y Ren, M Vigo, K van Deemter… – Language and …, 2015 – aclweb.org … systems. NLG systems are frequently based on a pipeline such as: a) content selection b) aggregation c) grouping and ordering Following the NLG pipeline, after content selection, we must consider grouping and ordering. Often … Related articles All 6 versions

TBI-Doc: Generating Patient & Clinician Reports from Brain Imaging Data P Jordan, NL Green, C Thomas, S Holm – INLG 2014, 2014 – aclweb.org … The architecture of TBI-Doc (shown by the re- mainder of Figure 3) follows the standard NLG pipeline (Reiter et al., 2000) and is similar to the architecture of the healthcare-related systems de- scribed in (Green et al., 2011; Hunter et al., 2012; Scott et al., 2013). … Related articles All 9 versions

A framework for creating natural language descriptions of video streams MUG Khan, N Al Harbi, Y Gotoh – Information Sciences, 2015 – Elsevier … First column in Fig. 4 presents three main tasks of the NLG pipeline, while second column shows subtasks against each main task. Third column presents corresponding tasks for natural language description of visual images. … Related articles All 3 versions

Narrative variations in a virtual storyteller SM Lukin, MA Walker – Intelligent Virtual Agents, 2015 – Springer … structures [ 14 , 21 ]. Fig. 5. NLG pipeline method of the ES Translator. dsynts are dependency structures where the nodes are labeled with lexemes and the arcs of the tree are labeled with syntactic relations. The dsynts formalism … Cited by 1

Explanation architecture for companion systems F Nothdurft, W Minker – Applied Electrical Engineering and …, 2011 – ieeexplore.ieee.org … user interface. Text generation is done by using a typical natural lan- guage generation (NLG) pipeline [20]. The first step in a NLG-pipeline, namely the content determination is realized by the explanation machine. The selected … Related articles All 2 versions

Latent User Models for Online River Information Tailoring X Han, S Sripada, KCJA Macleod, AAR Ioris – INLG 2014, 2014 – aclweb.org … visiting times of all river stations. 4 NLG for Online Tailoring Our NLG pipeline system takes numeric data of a given river station as input, and outputs a tai- lored description for that river station. The sys- tem analyzes data of … Cited by 1 Related articles All 8 versions

Presenting query results in NL for Quelo. XI Juárez-Castro – loria.fr Page 1. Presenting query results in NL for Quelo. X. Isabel Juárez-Castro Free University of Bozen-Bolzano 1 Introduction The Semantic Web has made available a huge amount of information stored in ontologies. The usual … Related articles

Challenges of Military Applications for Data-to-Text Generation H Hastie – researchgate.net Page 1. Challenges of Military Applications for Data-to-Text Generation Helen Hastie School of Mathematical and Computer Sciences Heriot-Watt University, Edinburgh, UK h.hastie@hw.ac.uk Abstract This position paper describes …

Natural language generation through case-based text modification J Valls, S Ontañón – Case-Based Reasoning Research and Development, 2012 – Springer … Surface realization is the final stage in the NLG pipeline [11], and is in charge of generating actual natural language text given a sentence specification, embodying all decisions related to the grammar and morphology of a particular language [6]. We present a case-based … Cited by 6 Related articles All 4 versions

Extracting Surface Realisation Templates from Corpora. TD Tadeu, EM de Novais, I Paraboni – LREC, 2010 – hnk.ffzg.hr … Briefly, surface realisation concerns the production of surface strings from a given semantic representation, and it is considered one of the final stages in the language generation process in a standard NLG pipeline architecture (Reiter, 2007.) … Related articles All 10 versions

An intelligent query interface based on ontology navigation E Franconi, P Guagliardo, M Trevisan – … on Visual Interfaces to the Social …, 2010 – Citeseer Page 1. An intelligent query interface based on ontology navigation Enrico Franconi franconi@inf.unibz.it Paolo Guagliardo paolo.guagliardo@stud-inf.unibz.it Marco Trevisan marco.trevisan@stud-inf.unibz.it KRDB Research Centre Free University of Bozen-Bolzano … Cited by 23 Related articles All 5 versions

Experiences with planning for natural language generation A Koller, R Petrick – Computational Intelligence, 2011 – Wiley Online Library … The chain of domain planning, sentence planning, and surface realization is sometimes called the “NLG pipeline” (Reiter and Dale 2000). Viewing generation as a planning problem has a long tradition in the NLG literature. … Cited by 40 Related articles All 14 versions

Does Personality Matter? Expressive Generation for Dialogue Interaction MA Walker, J Sawyer, G Lin, S Wing – Natural Interaction with Robots, …, 2014 – Springer … Our work on expressive and adaptive generation for dialogue interaction starts with the standard NLG pipeline architecture [13,22,32,45,47,51]. Figure 26.2 shows how we integrated the standard NLG pipeline architecture into AT&T Communicator [53, 59, 60]. … Related articles All 3 versions

Gen-Meta: Generating metaphors by combining AI and corpus-based modeling A Gargett, J Barnden – Web Intelligence, 2015 – content.iospress.com … At this point, the missing ECG generation component presents a gap in our NLG pipeline, to be filled by the DS generation mechanism, by encoding the Goal Schema as a Goal Tree, then proceeding to realise this content as an appropriate surface string. …

Improved text generation using n-gram statistics EM De Novais, TD Tadeu, I Paraboni – Advances in Artificial Intelligence– …, 2010 – Springer … In the final stages of the NLG pipeline, sentence spec- ifications are assigned content words and then linearised, in the task known as surface realisation [2,3]. Surface realisation takes as an input a semantic representation of what to say, and determines how to say it, producing … Cited by 4 Related articles All 8 versions

Partial-tree linearization: generalized word ordering for text synthesis Y Zhang – Proceedings of the Twenty-Third international joint …, 2013 – dl.acm.org … By allowing POS and dependency relations to be associated with any subset of in- put words, partial-tree linearization is more prac- tical for a dependency-based NLG pipeline, such as transfer-based MT and abstractive text summariza- tion. … Cited by 6 Related articles All 5 versions

Query2Question: Translating Visualization Interaction into Natural Language C Weaver – ieeexplore.ieee.org … Online generation is the dynamic generation of questions as a user interacts with a visualization. The relationship table populated by offline generation is given as input to online generation. The Q2Q translation architecture is based on the well- known NLG Pipeline Method. … Related articles All 3 versions

Sentence Generation: Input, Algorithms and Applications C Gardent – Actes des conférences TALN 2011 et Recital 2011 – webloria.loria.fr … at the end of the Natural Language Generation (NLG) pipeline, to produce the clauses making up the generated text in experiments on symbolic, transfer-based machine translation (http://www.emmtee.net/) to develop intelligent, user friendly interfaces to knowledge bases (cf. … All 12 versions

Generating Sentence Planning Variations for Story Telling SM Lukin, LI Reed, MA Walker – 16th Annual Meeting of the Special …, 2015 – aclweb.org … made with complex sen- tences in stories. We sum up and discuss future work in Sec. 5. 2 Story Generation Framework Figure 1: NLG pipeline method of the ES Trans- lator. Fig. 1 illustrates our overall architecture, which uses …

System personality and adaptivity in affective human-computer interaction S Konstantopoulos, V Karkaletsis – International Journal on Artificial …, 2013 – World Scientific … suited for such situations, where linguistic templates with empty information slots are filled with facts from the machine-readable information store to produce textual descrip- tions.4 Adaptivity was, then, restricted to the content selection phase of the NLG pipeline, and, more … Cited by 1 Related articles All 7 versions

Charting the potential of description logic for the generation of referring expressions Y Ren, K Van Deemter, JZ Pan – Proceedings of the 6th International …, 2010 – dl.acm.org … many researchers in focussing on the semantic core of the GRE problem: we shall generate descriptions of semantic content, leaving the decision of what words to use for ex- pressing this content (eg, ‘the ancient dog’, or ‘the dog which is old’) to later stages in the NLG pipeline. … Cited by 8 Related articles All 13 versions

RTG Based Surface Realization from Dependency Representation S Narayan – 2011 – homepages.inf.ed.ac.uk … 56 v Page 8. List of Figures 2.1 NLG Pipeline . . . . . … The architecture of an NLG system may be viewed as a pipeline of various tasks like Content Determination, Document Planning, and Surface Realization. Surface real- ization is the back-end task in the NLG pipeline. … Cited by 2 Related articles All 2 versions

Generation of landmark-based navigation instructions from open-source data M Dräger, A Koller – Proceedings of the 13th Conference of the …, 2012 – dl.acm.org … Below, we describe the individual components of the system. We mostly follow a standard NLG pipeline (Reiter and Dale, 2000), with a focus on the sentence planner and an extension to interac- tive real-time NLG. 760 Page 5. … Cited by 8 Related articles All 14 versions

Automating Direct Speech Variations in Stories and Games SM Lukin, JO Ryan, MA Walker – Tenth Artificial Intelligence and …, 2014 – users.soe.ucsc.edu … and Walker 2011; Cavazza and Charles 2005; Rowe, Ha, and Lester 2008). Figure 1: NLG pipeline method of the ES Translator. (Walker et al. 2013) explore using a dynamic and cus- tomizable NLG engine called PERSONAGE … Cited by 1 Related articles All 2 versions

Generating Referring Expressions with OWL2 Y Ren, K Van Deemter, JZ Pan – 23rd International Workshop on …, 2010 – dl.kr.org … core of the GRE problem: we shall be content to generate descriptions of logical/semantic content, leaving the decision of what English (or other languages) words to use for expressing this content (eg,’the ancient dog’, or ‘the dog which is old’) to later stages in the NLG pipeline. … Related articles All 10 versions

Stochastic language generation in dialogue using factored language models F Mairesse, S Young – Computational Linguistics, 2014 – MIT Press Cited by 7 Related articles All 14 versions

Adaptive explanation architecture for maintaining human-computer trust F Nothdurft, G Bertrand, H Lang… – Computer Software and …, 2012 – ieeexplore.ieee.org … With respect to a common natural language generation (NLG) pipeline [19], explanation machines are part of content planning or content determination because they select the appropriate ingredients for the explanation, though on a very abstract level. … Cited by 6 Related articles All 5 versions

Data-to-text summarisation of patient records: Using computer-generated summaries to access patient histories D Scott, C Hallett, R Fettiplace – Patient education and counseling, 2013 – Elsevier … 2.1.3. Architecture. The system design of the Report Generator follows a classical NLG pipeline architecture, with a Content Selector, MicroPlanner and Syntactic Realiser [24]. These roughly correspond to deciding what to say, how to say it and then actually saying it. … Cited by 8 Related articles All 10 versions

Integrating natural language generation and model-based reasoning for explanation generation D Matheson, GM Coghill… – … Intelligence (UKCI), 2012 …, 2012 – ieeexplore.ieee.org … The language generation process is modelled on the standard NLG pipeline architecture [22] which requires three stages: • Content determination, where the basic document plan is made and populated with messages that fulfil the communicative goal of the text … Cited by 1 Related articles All 2 versions

A Decision Theoretic Approach to Natural Language Generation N McKinley – 2014 – etd.ohiolink.edu Page 1. A DECISION THEORETIC APPROACH TO NATURAL LANGUAGE GENERATION by NATHAN MCKINLEY Submitted in partial fulfillment of the requirements For the degree of Master of Science Thesis Advisor: Dr. Soumya Ray … Cited by 2 Related articles All 10 versions

A Simple Surface Realization Engine for Telugu SRS Dokkara, SV Penumathsa, SG Sripada – ENLG 2015, 2015 – anthology.aclweb.org … Surface realization is the final subtask of an NLG pipeline (Reiter and Dale, 2000) that is responsi- ble for mechanically applying all the linguistic choices made by upstream subtasks (such as mi- croplanning) to generate a grammatically valid surface form. …

Quelo natural language interface: Generating queries and answer descriptions E Franconi, C Gardent… – … Interfaces for Web …, 2014 – hal.archives-ouvertes.fr … 3.3 Overall natural language generation architecture Starting from a query tree (Section 3.1) or query tree plus set of selected nodes, the generator follows a traditional NLG pipeline sequencing the following mod- ules. Document planning. … Cited by 2 Related articles All 5 versions

[BOOK] Multilingual text generation from structured formal representations D Dannélls – 2013 – gupea.ub.gu.se Page 1. Dana Dannélls Multilingual text generation from structured formal representations Page 2. Data linguistica <http://www.svenska.gu.se/publikationer/data-linguistica/> Editor: Lars Borin Språkbanken Department of Swedish University of Gothenburg 23 • 2012 Page 3. … Related articles All 5 versions

Automatic generation of human-like route descriptions: a corpus-driven approach R Teles, B Barroso, A Guimaraes… – Journal of Emerging …, 2013 – ojs.academypublisher.com … Figure 1. NLG system example The final text of an NLG pipeline should meet the following features: (1) be linguistically correct, (2) clearly convey the information input, (3) respond to the proposal 414 JOURNAL OF EMERGING TECHNOLOGIES IN WEB INTELLIGENCE, VOL. … Cited by 3 Related articles All 4 versions

Discriminative Syntax-Based Word Ordering for Text Generation Y Zhang, S Clark – Computational Linguistics, 2015 – MIT Press … We call this word ordering task the partial-tree linearization problem, a task that is perhaps closer to NLP applications than both the abstract word ordering task and the full tree linearization problem, in the sense that NLG pipelines might provide some syntactic relations between …

Generation of German narrative probability exercises R Boer Rookhuiszen – 2011 – essay.utwente.nl … 64 5.1.1 The NLG pipeline in Genpex . . . . . … 2.2.1 The NLG processes in the Narrator The Natural Language Generation (NLG) processes in this Narrator are mainly based on the general NLG pipeline as proposed by Reiter and Dale [RD97]. … Related articles All 4 versions

Generating and Interpreting Referring Expressions in Context DA Smith – 2013 – alumni.media.mit.edu Page 1. Generating and Interpreting Referring Expressions in Context by Dustin Arthur Smith BS, Wake Forest University (2005) SM, Massachusetts Institute of Technology (2007) Submitted to the Program of Media Arts and Sciences, School of Architecture and Planning … Related articles All 2 versions

Generating numerical approximations R Power, S Williams – Computational Linguistics, 2012 – MIT Press … 7 In terms of the standard NLG pipeline architecture (Reiter 1994), this would mean that mathematical form is already decided in the phase of content determination, so that the distinction between fractions, ratios, and percentages is an input to the subsequent phases (sentence … Cited by 9 Related articles All 11 versions

Generation of Natural Language Texts from Process Models H Leopold – Natural Language in Business Process Models, 2013 – Springer … Subsequently, in the sections 5.2.2 to 5.2.7, we introduce each component in detail. 5.2.1 Overview The architecture of our text generation approach is building on the traditional NLG pipeline concept [280]. The basic rationale of the approach is to utilize the existing Page 7. …

The generation of natural descriptions: corpus-based investigations of referring expressions in visual domains HAE Viethen – 2011 – researchonline.mq.edu.au … The pipeline is pictured in Figure 2.1. It consists of three stages, Document Planning, Microplanning and Realisation, each with its own subtasks. In this view, reg is one of the tasks of the pipeline taking place during the Microplanning stage of the nlg pipeline. … Cited by 9 Related articles All 2 versions

Learning to tell tales: automatic story generation from Corpora ND McIntyre – 2011 – era.lib.ed.ac.uk … Page 24. 6 Chapter 1. Introduction research which focuses on individual components, we propose an end-to-end system realising the whole NLG pipeline. The evaluation methods developed in … to partially implement the NLG pipeline as described in Reiter and Dale (2000), … Cited by 2 Related articles All 4 versions

Computational style processing F Khosmood – Ph. D. Dissertation, University of California Santa …, 2011 – works.bepress.com Page 1. UNIVERSITY OF CALIFORNIA SANTA CRUZ COMPUTATIONAL STYLE PROCESSING A dissertation submitted in partial satisfaction of the requirements for the degree of DOCTOR OF PHILOSOPHY in COMPUTER SCIENCE by Foaad Khosmood December 2011 … Cited by 2 Related articles All 2 versions

On Applying Controlled Natural Languages for Ontology Authoring and Semantic Annotation BP Davis – 2013 – vmserver14.nuigalway.ie Page 1. On Applying Controlled Natural Languages for Ontology Authoring and Semantic Annotation Brian Patrick Davis Submitted in fulfillment of the requirements for the degree of Doctor of Philosophy PRIMARY SUPERVISOR: Prof. … Cited by 1 Related articles All 3 versions

[BOOK] Natural Language Generation in Interactive Systems A Stent, S Bangalore – 2014 – books.google.com Page 1. Natural Language Generation in Interactive Systems tunea by Amanda Stel’ltand Srinivas Bangalore Page 2. Natural Language Generation in Interactive Systems An informative and comprehensive overview of the state … Cited by 2 Related articles All 3 versions

Accident modelling and analysis in process industries A Al-Shanini, A Ahmad, F Khan – Journal of Loss Prevention in the Process …, 2014 – Elsevier … Beale, 2006). 3, July 30, 2004, Ghislenghien, Belgium, Fireball and VCE, Release due to pipe damage by unknown reason, NLG pipeline, 24 deaths, over 120 injuries and property damage, (Mahgerefteh and Atti, 2004). 4, March … Cited by 6 Related articles All 5 versions

Controlling user perceptions of linguistic style: Trainable generation of personality traits F Mairesse, MA Walker – Computational Linguistics, 2011 – dl.acm.org … The PERSONAGE architecture (Figure 1) builds on a standard natural language generation (NLG) pipeline architecture as described in Reiter and Dale (2000), Kittredge, Korelsky, and Rambow (1991), and Walker and Rambow (2002). … Cited by 48 Related articles All 16 versions

Report Generation L Wanner – HANDBOOK OF – nozdr.ru Page 559. 22 Report Generation Leo Wanner Institució Catalana de Recerca i Estudis Avançats and Universitat Pompeu Fabra 22.1 Introduction….. 533 22.2 What Makes Report Generation a Distinct Task?…. … Related articles All 2 versions

Modelling Incremental Self-Repair Processing in Dialogue. J Hough – 2014 – qmro.qmul.ac.uk Page 1. 1 Modelling Incremental Self-Repair Processing in Dialogue Julian Hough Submitted for the degree of Doctor of Philosophy Queen Mary University of London 2015 Page 2. 2 Declaration I hereby declare that the work … Cited by 1

Evaluating the Perception of Personality and Naturalness in Computer Generated Utterances C McCormick – 2012 – scss.tcd.ie Page 1. Evaluating the Perception of Personality and Naturalness in Computer Generated Utterances by Christopher McCormick, B.Sc.(Hons) Dissertation Presented to the University of Dublin, Trinity College in fulfillment of the requirements for the Degree of … Related articles All 3 versions

Referring expression generation towards mediating shared perceptual basis in situated dialogue R Fang – 2014 – cse.msu.edu … However, when we consider REG in isolation, it quickly becomes clear that things are more complicated than this. It appears that some of the other subtasks within the NLG pipeline are in actual fact tasks that have to be tackled within REG itself as well. A …

Natural language processing methods for attitudinal near-synonymy ME Gardiner – 2013 – researchonline.mq.edu.au Page 1. Natural Language Processing Methods for Attitudinal Near-Synonymy Mary Elizebeth Gardiner Bachelor of Science (Honours), Macquarie University, Sydney Bachelor of Arts, University of Sydney This dissertation is presented for the degree of Doctor of Philosophy at … Related articles All 4 versions

Extensions to PKS in support of dialogue and conversational acts R Petrick, M Steedman – i61www.ira.uka.de Page 1. 17/6-2010 Page 1 of 99 IST-FP6-IP-027657 / PACO-PLUS Last saved by: UEDIN Public Project no.: 027657 Project full title: Perception, Action & Cognition through learning of Object-Action Complexes Project Acronym: PACO-PLUS Deliverable no.: D5.1.3 … Related articles

Model for evaluating the global impact of smart grids in sub saharan Africa JF TREZZI, F SIGNA – 2014 – politesi.polimi.it … 25 1.2 Chinese population . . . . . 26 1.3 GDP China . . . . . 28 1.4 NLG pipelines China . . . . . 32 1.5 Wind power China . . . . . 34 1.6 Regional Grids . . . . . … Related articles All 2 versions