Micro-planning & Natural Language Generation

Notes:

Micro-planning is below the generation phase, aggregation and lexical choice. Micro-planning makes the lexical and syntactic transformation to structured information. Micro planning is responsible for lexical selection and ellipsis, a form of abstraction of filtered data. The latter part of the micro-planning phase is generating referring expressions. Thematization plays a fundamental role in micro-planning tasks. There is some overlap between micro-planning and realization. Realization is essentially responsible for ordering phrases in the output and also performing punctuation tasks.

A micro-planner produces a text specification, a post-processor step. The micro-planner solves the problem of selecting the content words by performing lexicalization and referencing, as well as aggregation of paragraphs and sentences. A micro-planner may also be known as a linguistic realizer. A surface realizer applies grammar rules to convert an abstract representation into text.

Micro-planning may also be known as sentence planning. Whereas, macro-planning comprises content selection and document structuring. Document planning which selects the most important events from earlier stages and groups them into a tree of linked events. Micro-planning and realization then translate this tree into coherent text, which collectively could be considered an “interpretation module”.

Data-to-text generation: This process involves translating structured data, such as databases or graphs, into coherent, human-readable text. It’s crucial in domains where human interpretation of raw data is challenging, facilitating understanding through narrative summaries, reports, or descriptions.
Data-to-text generation system: A sophisticated system designed to automate the transformation of structured data into human-readable narratives. Such systems typically use Natural Language Generation (NLG) techniques, ensuring the produced text maintains accuracy, clarity, and relevance.
Macro-planning: Within the realm of Natural Language Generation (NLG), macro-planning refers to the overarching strategy or blueprint for content organization. It determines the high-level structure of the generated text, deciding which information pieces should be included and their sequential order, ensuring logical coherence.
Micro-planner: A vital component in NLG systems, the micro-planner makes intricate linguistic decisions. These can range from word selection based on context, determining appropriate referring expressions to avoid ambiguity, or deciding when and how to aggregate similar pieces of information for brevity and clarity.
Micro-planners: Refers to multiple modules or entities within an NLG system that are responsible for the detailed linguistic choices that micro-planning entails. These entities collectively ensure the text’s fluency and appropriateness.
NLG micro-planner: This is a micro-planner specifically tailored for tasks in the domain of Natural Language Generation. It’s designed to make decisions that align with the intricacies and nuances of human language, ensuring the generated text is both linguistically and contextually appropriate.
NLG micro-planners: The collective term for multiple NLG micro-planners operating within a system, each potentially responsible for specific linguistic tasks or decisions.
NLG micro-planning: Within NLG systems, this is the phase where granular linguistic decisions are made. The aim is to ensure that the generated text is fluent, contextually relevant, and linguistically correct, often encompassing decisions about terminology, syntax, and aggregation.
Semantic clustering: This technique groups data or concepts based on their intrinsic meaning or semantic relationships rather than superficial characteristics. It’s essential in contexts like topic modeling, information retrieval, or when trying to discover underlying themes within large datasets.
Sentence planner: Within NLG systems, this component focuses on organizing and structuring information at the individual sentence level. It decides the internal structure of sentences, determining how different concepts or data points will be integrated cohesively.
Sentence planning: This process, pivotal in NLG, concerns the organization of information within individual sentences. It focuses on ensuring that each sentence is structured logically, clearly conveys its intended message, and fits within the broader narrative.
Sentence realization: In NLG, this is the phase where abstract sentence structures or templates are transformed into actual textual strings. It’s essentially the act of “realizing” or “materializing” a sentence plan into tangible text.
Surface realizer: Operating at the final stage of NLG, the surface realizer takes abstract sentence representations and translates them into the final text. This involves handling language-specific tasks like determining correct word order, applying grammatical rules, and ensuring appropriate inflections based on context.
Tactic planning: In the context of NLG, tactic planning dives into the finer details of text representation. It emphasizes specific strategies and tactics for representing information, ensuring the text is engaging, accurate, and suits the target audience.
Text generator: A tool or system that, often using advanced algorithms or AI, produces text. While the source can be non-textual data, the output is human-readable content, whether it’s a news article, a product description, or any other form of narrative.
Wide-coverage micro-planners: These micro-planners have a broad operational range, capable of handling a diverse set of linguistic constructions and contexts. Unlike domain-specific micro-planners, they’re equipped to work in varied scenarios, making them versatile in different NLG tasks.

References:

Integrated Models for Information Communication Systems and Networks: Design and Development (2013)

A statistical nlg framework for aggregated planning and realization
R Kondadadi, B Howald, F Schilder – … of the 51st Annual Meeting of the …, 2013 – aclweb.org
… Our system does consolidate many traditional components (macro- and micro-planning, lexical choice and … The quality and the linguistic variability of the generated text is di- rectly … also presented a new automatic metric to evaluate generated texts at document collection level to …

Creating training corpora for nlg micro-planners
C Gardent, A Shimorina, S Narayan… – Proceedings of the 55th …, 2017 – aclweb.org
… To train Natural Language Generation (NLG) sys- tems, various input-text corpora have been devel- oped which associate (numerical, formal, linguis … focus on how to create data- to-text corpora which can support the learning of micro-planners ie, data-to-text generation sys …

Creating training corpora for nlg micro-planning
C Gardent, A Shimorina, S Narayan… – 55th annual meeting of …, 2017 – hal.inria.fr
… guistic Annotations and (iii) crowdsourced bench- marks1. In this paper, we focus on how to create data- to-text corpora which can support the learning of wide-coverage micro-planners ie, generation sys- tems that handles such NLG subtasks as lexicali …

Domain adaptable semantic clustering in statistical nlg
B Howald, R Kondadadi, F Schilder – Proceedings of the 10th …, 2013 – aclweb.org
… summarize NLG as follows: (1) Macro Planning creating a document plan; (2) Micro Planning sentence planning; (3 … Surface Realization concatenating the information from (1- 2) into coherent and grammatical text; and (4) Physical Presentation document layout considerations …

MIME-NLG in Pre-hospital Care
A Schneider, A Mort, C Mellish, E Reiter… – … Language Generation, 2013 – aclweb.org
… Events that will be mentioned in the text are selected and structured into a list … In the micro-planning step the structure of the document plan is linearised and sentences are … Whereas some parts of the handover document (eg patient demographics) are relatively stylised, the main …

Gennext: A consolidated domain adaptable nlg system
F Schilder, B Howald, R Kondadadi – … on Natural Language Generation, 2013 – aclweb.org
… There is often a structured database underlying the domains of corpus text, the fields of which are used for domain specific entity tag- ging (in addition to domain general entity tagging [eg … 2013) – semantic clustering and micro-planning and Kondadadi … 2013) – document planning …

Natural language generation and semantic technologies
K Staykova – Cybernetics and Information Technologies, 2014 – degruyter.com
… of the whole text and is often used as a specification of the discourse structure of the text. The second module is Sentence Planner (known also as Micro-planner or Tactic planning … It deals with mapping of the document plan to linguistic structures and groups the information into …

Natural language generation in the context of the semantic web
N Bouayad-Agha, G Casamayor, L Wanner – Semantic Web, 2014 – content.iospress.com
… also referred to as macro- planning), 2) sentence planning (otherwise known as micro-planning in opposition … The sentence planner is in charge of mapping the re- sulting text plan to the linguistic … It is therefore obvious that it is the document planning module and, to some extent …

Enabling text readability awareness during the micro planning phase of NLG applications
P Moraes, K McCoy, S Carberry – … Natural Language Generation …, 2016 – aclweb.org
… in Enabling text readability awareness during the micro planning phase of NLG applications1 Priscilla … 121 Page 2. 2 order to convey a message or to make a point re- garding the topic being discussed in the document or article. To …

Towards diet management with automatic reasoning and persuasive natural language generation
L Anselma, A Mazzei – Portuguese Conference on Artificial Intelligence, 2015 – Springer
… say, that is which information contents will be communicated; (2) In the micro- planning, the focus … the contents of information that have to be communicated, ie the document planning, are … Aarts, EHL: Adaptive persua- sive systems: A study of tailored persuasive text messages to …

Planning for Natural Language Generation in GF
G Lobanov, K Angelov – Workshop on Logic and Algorithms in …, 2017 – diva-portal.org
… what makes the final applications multilingual: all steps prior to the micro planning, eg document … format from the weather forecasting cloud service DarkSky1, and builds a document plan … of the weather report grammar is modular, reflecting the compositional structure of the text …

Predicting sentential semantic compatibility for aggregation in text-to-text generation
V Chenal, JCK Cheung – Proceedings of COLING 2016, the 26th …, 2016 – aclweb.org
… Aggregation is an important step in micro-planning in the traditional data-to-text NLG pipeline … fragments are therefore very close to clauses, and a sentence in a text can contain … within a same paragraph and 20 pairs originated from different paragraphs within a same document …

Recent Advances in Natural Language Generation: A Survey and Classification of the Empirical Literature
R Perera, P Nand – Computing and Informatics, 2017 – cai.sk
… the document planner is to generate a structured message which can be fed to the micro-planner … Firstly, SUMTIME-MOUSAM feed time series numerical data to document planner unit … and interest- ingly, the realization engine used here does not output surface text as expected …

Hypertext Summarization for Hotel Review
M Tien, F Portet, C Labbé – 2015 – hal.archives-ouvertes.fr
… 2.3 Document Planning / Micro Planning … as a small hyper- text body code in order to be able to represent it better(with bold text, extra information in … quality of SuRe system base on the us- ing a set of characteristic that were defined for the Document understanding conference …

Communication Mediated through Natural Language Generation in Big Data Environments: The Case of Nomao
JS Vayre, E Delpech, A Dufresne… – Journal of Computer and …, 2017 – file.scirp.org
… The output of the micro-planning stage is a sequence of lemmas (non-inflected words) which should be … the task; see the essential load); – the effort required to navigate inside the document(ie, to … text 1 is less complex than text 2. Text 1 has 11 sentences (s) whereas text 2 has …

Creating Training Corpora for NLG Micro-Planning
CGA Shimorina, SNL Perez-Beltrachini – pdfs.semanticscholar.org
… To train Natural Language Generation (NLG) sys- tems, various input-text corpora have been devel- oped which associate (numerical, formal, linguis … focus on how to create data- to-text corpora which can support the learning of micro-planners ie, data-to-text generation sys …

An automatic health surveillance chart interpretation system based on Indonesian language
I Aulia, AM Barmawi – Advanced Computer Science and …, 2015 – ieeexplore.ieee.org
… These parts are integrated by the NLG architecture suggested by Reiter and Dale in [3], so that Data-to-Text architecture consists of Signal Analysis, Data Interpretation, Document Planning, and Micro Planning and Realisation [6] as shown in Fig. 2. ^ / DW Z W W DZ ^ D …

Natural Language Generation
O Dušek – 2013 – ufal.mff.cuni.cz
… 1. Introduction: what is NLG? a) Textbook NLG pipeline b) How real systems differ 2. Examples of real NLG systems (many) • different stages of NLG 2/ 40 Ond?ej Dušek Natural Language Generation Page 7 …

Evaluative Pattern Extraction for Automated Text Generation
CC Lee, SK Hsieh – … the 9th International Natural Language Generation …, 2016 – aclweb.org
… patterns of evaluative language agglomerative structure would facilitate both the macro/ micro planning in NLG … extraction algorithms and for text summari- zation in the process of document planning in … For text generation, pattern is a significant feature as a representation of the …

Constructing Sentences from Text Fragments: Aggregation in Text-to-text Generation
V Chenal – 2017 – digitool.library.mcgill.ca
… that framework. Specifically, we focus on aggregation, an important step in micro-planning in the tra … based systems. It combines document and sentence planning and surface realization into … To generate text for a given domain, the system runs input data through a statistical …

Natural Language Generation
G Godbole – oaji.net
… an advanced model consisting of three major steps- Document planner, Micro-planner and Surface … They are templates which explicitly specify sentence ordering and grouping for a document … date with significant modifications in the general steps of the method to generate text …

WiseReporter: A Korean Report Generation System
Y Noh, SJ Choi, SB Park, SY Park – Proceedings of the IJCNLP 2017 …, 2017 – aclweb.org
… WiseReporter adopts a pipelined architecture for natural language generation following several studies on ontology … to as macro-planning), (ii) sen- tence planning (also known as micro-planning), and (iii … sponsible for mapping the text plan to a linguis- tic structure, grouping …

Language Generation from DB Query
K Kocijan, B Bekavac, K Šojat – … Automatic Natural Language Generation …, 2017 – aclweb.org
… Macro-plan- ning comprises content selection and document structuring, whereas micro-planning usually refers to the … Apart from morphological mod- ifications, this last stage also includes typographical adjustment and formatting and provides the final form of the text …

Scalable Natural Language Report Management Using Distributed IE and NLG from Ontology
M Madhavan – 2013 – academia.edu
… 42 8.1.4 Micro Planning … NLP. Recent activities in multimedia document processing like automatic annotation and … The over- all goal being to create a more easily machine-readable text to process the sentences. Typical subtasks of IE include [6] …

BahaBa: A Route Generator System for Mobile Devices
RV Regalado, MB Haw, MA Martinez… – Proceedings of the …, 2013 – aclweb.org
… Micro planning makes a list of sentences and identifies what information to be used for the route to be undertaken … Text Plan … But since we are using template-based NLG, it can easily be adapted to other languages by simply translating the templates that are stored in the …

Automatic text generation by learning from literary structures
A Daza, H Calvo, J Figueroa-Nazuno – Proceedings of the Fifth …, 2016 – aclweb.org
… Instead of generating language fol- lowing a detailed document plan, we propose to start with … of the diversity of sentences that can be produced on the micro-planning step, which … the importance of having a well-structured knowledge base for optimizing the text generation layer …

JSREAL: A text realizer for web programming
N Daoust, G Lapalme – Language Production, Cognition, and the Lexicon, 2015 – Springer
… 2000), NLG involves three stages: macro-planning (content choice and organization), micro-planning (abstract specification … used standalone for linguistic demonstrations or be integrated into complex text generation projects … of JSreal are similar to those of SimpleNLG: they …

Generating récit from sensor data: evaluation of a task model for story planning and preliminary experiments with GPS data
BAB Miranda, S Caffiau, C Garbay… – … Language Generation …, 2015 – aclweb.org
… To go a step further, a basic NLG system to generate narrative from activities extracted from GPS … One of the most important task for the text generation part is to adopt a more … We are working on re-implementing the micro-planner used in the BabyTalk project (Portet et al., 2009) …

The WebNLG Challenge: Generating Text from RDF Data
C Gardent, A Shimorina, S Narayan… – … Language Generation, 2017 – aclweb.org
… Previous Natural Language Generation (NLG) chal- lenges have focused on surface realisation (Banik et al., 2013; Belz et al … used a tem- plate or grammar-based pipeline framework with some NLG module: UTILBURG … rules mapping a triple (or a triple set) to a text observed in …

RDF2PT: Generating Brazilian Portuguese Texts from RDF Data
D Moussallem, TC Ferreira, M Zampieri… – arXiv preprint arXiv …, 2018 – arxiv.org
… A generic NLG pipeline is composed by three tasks which are Document Planing, Micro Planning and Realization … 3.3. Document Planning This initial phase is divided into two sub-tasks … for deciding what information a certain NLG system should include in the generated text …

Natural language generation library for Spanish
S García Méndez – 2017 – pequod.det.uvigo.es
… IV. Tools survey Text realisation is a NLG subtask which consists on creating a text in a human language from a syn- tactic representation … CRISP [17] is a NLG system that tries to cast both micro-planning and sentence realisation as an AI 3 planning problem …

Learning to summarize time series data
PKV Sowdaboina, S Chakraborti, S Sripada – … on Intelligent Text …, 2014 – Springer
… The process begins with document planning (also called content planning) which performs content selection, followed by micro-planning which analyses ways of expressing the information linguistically and lastly the surface realization of grammatical text …

Adapting graph summaries to the users’ reading levels
P Moraes, K McCoy, S Carberry – … Natural Language Generation …, 2014 – aclweb.org
… As stated by (Williams & Reiter, 2005b), most NLG systems generate text for readers with good … Their results show that, for these target readers, the micro planning choices made by SkillSum … Siddharthan, 2003) pro- poses a regeneration phase for syntactical text simplification in …

The Code2Text Challenge: Text Generation in Source Libraries
K Richardson, S Zarrieß, J Kuhn – … on Natural Language Generation, 2017 – aclweb.org
… 2016. The WebNLG Challenge: Generating Text from DBPedia Data … 2017. Creating Train- ing Corpora for NLG Micro-Planning. In Proceedings of ACL … 2017. Survey of the State of the Art in Natural Language Generation: Core tasks, applications and evaluation …

The Code2Text Challenge: Text Generation in Source Code Libraries
K Richardson, S Zarrieß, J Kuhn – arXiv preprint arXiv:1708.00098, 2017 – arxiv.org
… 2016. The WebNLG Challenge: Generating Text from DBPedia Data … 2017. Creating Train- ing Corpora for NLG Micro-Planning. In Proceedings of ACL … 2017. Survey of the State of the Art in Natural Language Generation: Core tasks, applications and evaluation …

Generation of linguistic advices for saving energy: architecture
G Trivino, D Sanchez-Valdes – … Conference on Theory and Practice of …, 2015 – Springer
… module that produces an instantiated Granular Linguistic Model of Phenomena, that we relates with the Reiter’s document plan. The Micro planner and the Surface realizer are implemented by using the Text Generator that produces an instance of the Text template …

Natural Language Generation for Declarative Process Models
L Ackermann, S Schönig, M Zeising… – Workshop on Enterprise …, 2015 – Springer
… Table 1. Challenges for generating text from declarative process models. Phase. Challenge. Related Work. Applicable. Document Planning. Information Extraction. Message Selection and Population … (\times \). Text Structuring … Micro-planning. Lexicalization. Linguistic Label Analysis …

Translating Italian to LIS in the rail stations
A Mazzei – … European Workshop on Natural Language Generation …, 2015 – aclweb.org
… station messages are produced in real-time by us- ing text-to-speech, and the textual … ATLAS a logic seman- tics was adopted, that was the input of an expert system micro-planner … generator is more complex and adopts the standard pipeline architecture of the NLG (Reiter and …

Textual summarization of events leading to health alerts
A Jain, JM Keller – Engineering in Medicine and Biology Society …, 2015 – ieeexplore.ieee.org
… It was concluded that the clinicians were more inclined towards the automatic generated text summaries as compared to … domain, this stage where the information is modified to sound more intuitive is called the document planning and micro-planning and realization [8 …

Toward automatic generation of linguistic advice for saving energy at home
P Conde-Clemente, JM Alonso, G Trivino – Soft Computing, 2018 – Springer
… In addition, the “document structuring” sub-task provides order and structure over the information to be conveyed. Micro planner It produces a text specification … some characterization of the rules of grammar to some more abstract representation in order to produce a text (saying it …

Generating natural language descriptions from OWL ontologies: the NaturalOWL system
I Androutsopoulos, G Lampouras, D Galanis – Journal of Artificial …, 2013 – jair.org
… Our system generates texts in three stages, document planning, micro-planning, and surface realization, discussed in the following sections … Document planning consists of content selection, where the system selects the information to convey, and text planning, where it …

Generating explanations for pro-active assistance from formal action descriptions
S Bader – International Joint Conference on Ambient Intelligence, 2013 – Springer
… Layer 2 (Micro-planner) performs the lexicalisation, referencing and aggregation of paragraphs and sentences … 3 (Surface realiser) applies grammar rules to convert an abstract representation into text. In our system described below, the document- and part of the micro-planning …

Claim synthesis via predicate recycling
Y Bilu, N Slonim – Proceedings of the 54th Annual Meeting of the …, 2016 – aclweb.org
… In an SNLG system, af- ter the macro-planning and micro-planning stages (see (Reiter and Dale, 2000)) are … In other words, while the classical NLG paradigm is to first se- lect the content and then realize … Addition- ally, if S is a set of text segments, define: sim(l, S) = maxl ?Ssim(l …

Mapping the field of algorithmic journalism
KN Dörr – Digital Journalism, 2016 – Taylor & Francis
… “Natural Language Generation.” In The Handbook of Computational Linguistics and Natural Language Processing, edited by … Moreover, they outline a functional typology and subsume text creation applications as a … on content creation applications on the basis of NLG and its …

Opinionated Natural Language Generation
T Cagan – 2016 – idc.ac.il
… Selecting the rel- evant information to convey and the high level document structure stems … While this is true in general, there is some overlap between micro-planning and realization … In templates we find some representation of readily available text – whole sentences or phrases …

Data-driven broad-coverage grammars for opinionated natural language generation (ONLG)
T Cagan, SL Frank, R Tsarfaty – Proceedings of the 55th Annual Meeting …, 2017 – aclweb.org
… In classical NLG terms, these grammars offer the opportunity for both micro-planning and surface … where tm weight(c) is the weight of topic c in the topic distribution at the document-agenda in- tersection, and word weight(c, li) is the weight of the lexical head word li within the …

Query2Question: translating visualization interaction into natural language
M Nafari, C Weaver – IEEE transactions on visualization and …, 2015 – ieeexplore.ieee.org
… Document planning determines the infor- mation that needs to be communicated in the text. Micro- planning determines how the selected information can be communicated linguistically. Realization is responsible for generating grammatically correct text …

Communicating Naturally
T Dancy – 2014 – archives.datapages.com
… DATA ANALYTICS MESSAGE POOL DOCUMENT PLANNER OVERALL DOCUMENT PLAN SENTENCE PLANS MICRO PLANNER LINGUISTIC REALIZER … A contextually-appropriate plan for the overall structure of the text …

Validating the INTERPRETOR Software Architecture for the Interpretation of Large and Noisy Data Sets
A Salatian – … Models for Information Communication Systems and …, 2013 – books.google.com
… then Microplan- ning and Realisation which translates this tree into coherent text for reporting … Document planning is responsible for selecting the ‘important’data points from the input data … Micro planning is responsible for lexical selection and ellipsis-this is a form of abstraction of …

2. Home Address
E Charniak – pdfs.semanticscholar.org
… “Entropy Rate Constancy in Text” (with Dmitri … “Natural Language Generation with Vocabulary Constraints” Proceedings of the Ninth Workshop on Innovative Use of NLP for … “Micro-Planner Reference Manual” (with GJ Sussman and T. Winograd), AI Memo 203A, MIT Artificial …

Generating subjective responses to opinionated articles in social media: an agenda-driven architecture and a Turing-like test
T Cagan, SL Frank, R Tsarfaty – Proceedings of the Joint Workshop on …, 2014 – aclweb.org
… Micro Planning (below, the generation phase): How are we going to say it … topic, the sentiment for the topic in the document, and the sentiment for that topic in the user agenda … ‘Source’ indicates whether the response is from the pre- sented text snippet or a random other …

Modeling content and structure for abstractive review summarization
S Gerani, G Carenini, RT Ng – Computer Speech & Language, 2016 – Elsevier
… receives the AHT as input and generates an abstractive summary by performing micro-planning and sentence … starts by applying a discourse parser to each review and obtaining a document level discourse … b)). After obtaining the DT, we remove all words from the text spans of …

Content selection as semantic-based ontology exploration
L Perez-Beltrachini, C Gardent, A Revuz… – … Language Generation …, 2016 – aclweb.org
… Our ultimate goal is the automatic creation of a corpus of KB queries for development and evaluation of NLG systems … We automatically cre- ated training datasets from several text and knowl … 2016. A Statistical, Grammar-Based Approach to Micro- Planning …

I Read the News Today, Oh Boy
T Veale, H Chen, G Li – International Conference on Distributed, Ambient …, 2017 – Springer
… two levels of planning: macro-planning (what is it I want to say?) and micro-planning (how do I … the lemmatized forms and their POS tags as features of the words in each document for the … Again the LDA model, derived from a combination of full text stories and recent news tweets …

Explanations engine for energy management systems in buildings
AA Alyafi, JP Guillbaud, P Reignier… – … Data Acquisition and …, 2017 – ieeexplore.ieee.org
… planning), makes a lexical and syntactic transformation to this structured information (micro-planning) and to … a high-level internal representation that presents deep general-purpose NLG module … the linearized arbores-cent expression of the tables and it produces a French text …

Harmonization of conflicting medical opinions using argumentation protocols and textual entailment-a case study on Parkinson disease
A Groza, MM Nagy – … and Processing (ICCP), 2016 IEEE 12th …, 2016 – ieeexplore.ieee.org
… ralOWL adopts a pipeline architecture, structuring the process into document planning, micro-planning and surface … In the document planning step, the engine selects the relevant information from the ontology and … Finally, the resulted text is computed through surface realization …

Natural language interaction with semantic web ontologies
G Lampouras – 2015 – nlp.cs.aueb.gr
… vi Page 7. CONTENTS vii 2.3.1 Document Planning … 32 2.3.2 Micro-planning … For example, our system can generate the following text from the OWL statement above, if the ontology has been annotated with domain- dependent linguistic resources discussed below …

Modeling the clarification potential of instructions: Predicting clarification requests and other reactions
L Benotti, P Blackburn – Computer Speech & Language, 2017 – Elsevier
… Roughly speaking, our inference framework takes as input sentences like (1) and explains how (2) can be generated: it indicates what kinds of knowledge need to be represented and … Micro-planning has previously been used for natural language generation by Koller and …

Towards Automatic Explaining Arti? cially Intelligent Behaviour
R Hagenaars – 2013 – dspace.library.uu.nl
… require less effort than techniques with an input more abstracted from the surface text … There are several different architectures for NLG systems, but the most fre- quently espoused is … Document Planner The Document Planner receives the communicative goals and is to specify …

Communication Aspects: How Landmarks Enrich the Communication Between Human and Machine
KF Richter, S Winter – Landmarks, 2014 – Springer
… 6.1. This kind of architecture is quite typical for NLG-motivated route planners and reflects to a … This text plan is the input for the micro-planning phase, which decides how to … The latter part of the micro-planning phase is generating referring expressions—including selection of the …

New types of computational perceptions: Linguistic descriptions in deforestation analysis
P Conde-Clemente, JM Alonso, ÉO Nunes… – Expert Systems With …, 2017 – Elsevier
… Firstly, the Document Planner decides what information should be communicated in the output text … Then, the Micro Planner solves the problem of selecting the content words (nouns, verbs … of grammar to some more abstract textual representation; thus, producing a text which is …

Abstractive summarization of product reviews using discourse structure
S Gerani, Y Mehdad, G Carenini, RT Ng… – Proceedings of the 2014 …, 2014 – aclweb.org
… of the last component which generates a natural language summary by apply- ing micro planning and sentence … for EDU2 and so is less important for real- izing the core meaning of the document. After obtaining the DTs, we remove all words from the text spans of each EDU …

Giving good directions: order of mention reflects visual salience
ADF Clarke, M Elsner, H Rohde – Frontiers in psychology, 2015 – frontiersin.org
… Thus, either model predicts that more visually salient objects are placed early in the sentence … distinguish four strategies for ordering the landmark relative to the anchor, which we illustrate with examples from our corpus (all referring to targets in Figure 1), with text describing the …

Split and rephrase
S Narayan, C Gardent, SB Cohen… – arXiv preprint arXiv …, 2017 – arxiv.org
… Pairing Using the semantic information given by WEBNLG RDF triples and the information about the number of sentences present in a WEBNLG text produced by the sentence seg- mentation step, we produce all items of the form <(MC,C),1(M1,T1)…(Mn,Tn)l> such that …

Sorry, i don’t speak SPARQL: translating SPARQL queries into natural language
AC Ngonga Ngomo, L Bühmann, C Unger… – Proceedings of the …, 2013 – dl.acm.org
… In terms of the standard model of natural language generation proposed by Reiter & Dale [24], our … play the role of the document planner, in particular carrying out the task of document structuring, while the postprocessor step corresponds to the micro-planner, with focus …

Surface Realisation from Knowledge Bases
B Gyawali – 2016 – hal.inria.fr
… Instead, we build a corpus from heterogeneous sources of text related to the domain of the knowledge base for … 2.3 NLG from Ontologies … d’une architecture en pipeline qui modélise la tâche de génération en trois tâches séquentielles: Planification de Document, Planification de …

RoboCOP: A Robotic Coach for Oral Presentations
H Trinh, R Asadi, D Edge, T Bickmore – … of the ACM on Interactive, Mobile …, 2017 – dl.acm.org
… In addition to the message types, we also identified text structure patterns, aggregation patterns as well as lexicalization options for each of the message type classes. This forms the basic knowledge source for the document structuring and micro-planning processes …

On applying controlled natural languages for ontology authoring and semantic annotation
BP Davis – 2013 – aran.library.nuigalway.ie
… 4See http://www.w3.org/TR/owl2-overview/,OWL 2 Web Ontology Language Document Overview W3C … (NLG), involves the generation of the CLOnE language from an existing ontology. The text generator and CLOnE authoring processes combine to form a RoundTrip Ontology …

A contribution to the ranking and description of classifications
G Sánchez Hernández – 2013 – upcommons.upc.edu
… 64 xv Page 16. Germán Sánchez-Hernández 4.3 Document planning … B.14 Groups of messages affected by rule A.5 . . . . . 159 B.15 Filtered messages that will be mentioned in the final text … 139 NLG natural language generation …

Textual Summarization of Data Using Linguistic Protoform Summaries
A Jain – 2015 – search.proquest.com
… [4] A. Gatt, F. Portet, E. Reiter, J. Hunter, S. Mahamood, W. Moncur, and S. Sripada, “From data to text in the neonatal intensive care unit: Using NLG technology for decision … 2595-2599. [6] E. Reiter and R. Dale, “Building Natural Language Generation Systems,” 2006 …

Generating natural language explanations for entailments in ontologies
T Nguyen – 2013 – oro.open.ac.uk
… I would like to thank other members of the SWAT (Semantic Web Authoring Tool) project and the NLG (Natural Language Generation) group at the Open University—in particular Donia Scott, Alan Rector, Allan Third, Fennie Liang, Svetlana Stoyanchev, and Brian …

From Logic to Language
V Basile – 2015 – hal.inria.fr
… Document prepared with LATEXand typeset with pdfLATEX … If we consider whatever formalism is used as input to a NLG component as a language in itself, then … of a statistical component for generation, some form of alignment between the generation input and the text must be …

Déjà Vu All Over Again
T Veale – computationalcreativity.net
… form the equivalent of a textual n-gram language model whose outputs are plots, not sentences … one prologue text, and end a story at any action that possesses at least one epilogue text … of planning: macro-planning (what is it that I want to say?) and micro- planning (so how do …

Systemic functional linguistics, cognitive linguistics and psycholinguistics: Opportunities for dialogue
CS Butler – Functions of language, 2013 – jbe-platform.com
… job is to structure the discourse as a whole, and a sentence planner dedicated to the task of specifying the internal structure of text sentences, which interact … discusses empirical psycholin- guistic evidence for the distinction between macro- and micro-planning, as well …

Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
R Barzilay, MY Kan – Proceedings of the 55th Annual Meeting of the …, 2017 – aclweb.org
… Yasuhide Miura, Motoki Taniguchi, Tomoki Taniguchi and Tomoko Ohkuma. Unifying Text, Metadata, and User Network Representations with a Neural Network for Geolocation Pre- diction … Abstractive Document Summarization with a Graph-Based Attentional Neural Model …

Computerized spatial language generation for object location
G Lara, A De Antonio, A Peña – Virtual Reality, 2016 – Springer
… It was written in Micro Planner and Lisp … image processing, semantic representation and generation of local expressions, and includes a natural language generation processor … The system includes a text–mode interface, which provides users with written instructions from the …

Generation of textual summaries at different target reading levels: summarizing line graphs for visually impaired users
PS Moraes – 2016 – search.proquest.com
… multimodal documents to convey a message that augments the message of the document or to … More details on the micro planning phase (aggregation and lexical choice) are provided in Chapter 7 … In order to generate text that describes an entity, it is essential to first identify the …

The Listening Word Processor
JERG Cumming – The Science of Writing: Theories, Methods …, 2013 – books.google.com
… 361 Page 379. 362 REECE 8: CUMMING an improvement in text quality. Where … session. Similarly, planning does not occur only at the beginning of a composition session; writers plan constantly throughout text production. Further …

Image summarisation: human action description from static images
E Tsironi – 2014 – sapientia.ualg.pt
… to extract common information by using a graph merging technique of the dependency graphs of the annotated sentences … Figure 50: The distribution of sentence clarity per assessor … and more specifically with the description of human actions depicted in an image in form of text …

HCI from a Discourse Perspective Area: Discourse and Interaction GRACE Deliverable 5.1
R Bod, M Dastani, R Scha, H Zeevat – cogsci.ed.ac.uk
… Each sentence of a text that is being read, or in general each utterance that is added to an ongoing discourse, is interpreted in a context to which the sentences preceding it have contributed. The meaning of most utterances …

Modelo computacional para la generación de indicaciones en la localización de objetos en entornos virtuales: aspectos espaciales y perceptivos
G Lara López – 2016 – oa.upm.es
Page 1. DEPARTAMENTO DE LENGUAJES, SISTEMAS INFORMÁTICOS E INGENIERÍA DEL SOFTWARE UNIVERSIDAD POLITÉCNICA DE MADRID ESCUELA TÉCNICA SUPERIOR DE INGENIEROS INFORMÁTICOS TESIS DOCTORAL …

Thematic Patterning in English and Spanish: Contrastive Annotation of a Bilingual Newspaper Corpus for Linguistic and Computational Applications
LM Gutiérrez – 2015 – eprints.ucm.es
… 4.1 Text Types … In NLG, for exam- ple, thematization plays a fundamental role in micro-planning tasks (Hovy & Lavid, 2007 … of high linguistic abstraction has been virtually unexplored in other computational areas such as Information Extraction (IE) and Document Classification (DC …

Handbook of cognitive linguistics
E Dabrowska, D Divjak – 2015 – books.google.com
… which is the most prominent element in sentences containing a complement clause (such as I know that she left). Langacker (1991) adopts the traditional analysis, according to which the main clause verb is the profile determinant, pointing out that the sentence designates “the …