The CLEAR.TEXT project (TED2021-130707B-I00) researches how natural language processing technology can support the authoring of accessible content in Spanish for people with cognitive disabilities. Our main objective is to research, implement, deploy, evaluate, and ultimately provide robust technologies for natural language processing to support the authoring of accessible Spanish content for public sector organisations (at local, regional and national level) that is intelligible to people with cognitive disability, thereby widening their inclusion and empowerment in Europe. It is expected to impact positively the quality of life of people with cognitive disabilities, facilitating their access to educational, vocational, cultural, and social opportunities in public sector organisations 1https://cleartext.gplsi.es/ 2https://cvnet.cpd.ua.es/curriculumbreve/grp/es/procesamiento-del-lenguaje-y-sistemas-de- informacion-(gplsi)/596
to widening their inclusion and empowerment in Europe.
People with cognitive disabilities have significant limi- With this hypothesis in mind, the CLEAR.TEXT tations in their intellectual functioning and/or may also project1 funded by the Spanish Government and the Eurolack the ability to adapt to everyday situations. In fact, pean Union and developed by the GPLSI research group2 individuals with cognitive impairment have spoken and of the University of Alicante. The project focuses on written word comprehension deficit that may include research, implement, deploy, evaluate, and ultimately misinterpretation of literal meanings and dificulty un- provide robust technologies for NLP to support the auderstanding complex instructions. They are confused thoring of accessible Spanish content for public sector by idioms, figures of speech, abstractions, uncommon organisations (at local, regional and national level) that words, and lack of precision. is intelligible to people with cognitive disability, thereby
At present, natural language processing (NLP) tech- widening their inclusion and empowerment in Europe. nologies are mature enough to provide a sound basis This, in turn, will improve the ability to access written for the development of components to automatically de- information for all people, thereby reducing the risk of tect and remove obstacles to reading comprehension and exclusion for those with cognitive disability. The project generate additional content to facilitate reading com- is expected to impact positively the quality of life of peoprehension. Hence, we start with the hypothesis that ple with cognitive disabilities, facilitating their access to research, development and deployment of NLP technol- educational, vocational, cultural, and social opportunities ogy can support the authoring of accessible content in in public sector organisations.
divided into the following specific objectives:
The methodology has been organised into six main modules whose tasks and results are interlinked, as shown in Figure 1. An additional transversal seventh module focusses on coordination tasks. • O1. To analyse the main comprehension obstacles posed by the language used in the web content of Spanish public sector organizations, such as ministries and other government agencies, for people with cognitive disabilities. • O2. To analysis of the needs of people with cog
nitive disabilities. • O3. To research and adapt the COMPENDIUM
System to the needs of public sector documentation. • O4. To research, implement, deploy, and ultimately provide robust technologies to support the processing of structural complexity. • O5. To research, implement, deploy, and ultimately provide robust technologies to support the processing of ambiguity in meaning. • O6. To research, implement, deploy, and ultimately provide a robust text simplification system oriented towards public administration documentation. • O7. To evaluate the simplification system. • O8. To promote and disseminate the research results obtained from the project through diferent national and international media –including well indexed journals, conferences, seminars, etc., as well as exploit the potential for transferring this technology to society.
Module A. Establishment and Analysis of User Requirements
This module is organized into two tasks which are detailed below. 3. Human Resources Task A.1. Analysis of the main language phenomena. The objective of this task is to analyse the language pheA multidisciplinary research team, comprising 5 special- nomena, which will be referred to as obstacles to reading ists from the Computer Science and 3 from the Linguis- comprehension, both obstacles caused by ambiguity in tics fields; in all 7 PhD holders and 1 technician. All meaning and obstacles caused by structural complexity, researchers, 5 women and 3 men, are members of the such us dificult words; morphologically, orthographiGPLSI research group. cally, and phonetically long or complex words; figura
The team has extensive experience spanning 30 years tive language such as metaphor and idioms; long and of NLP-focused technological research and development, syntactically complex sentences; or semantically highly and specifically concerning the requested project in the specialised/technical words, among others. areas of word sense disambiguation, anaphora resolution, Task A.2. User requirements. The objective of this task coreference, named entity, lexical and syntactic analysis, is to determine the users’ requirements. For this, it is text summarization, and text simplification -[ ? ], [? ], [? ifrst necessary to determine and analyze the types of ], [? ], [? ], [? ], [? ], [? ]. cognitive disabilities (dysphasia, Down syndrome, Fragile X syndrome and Autism Spectrum disorder) and, second, to establish user requirements and needs to facilitate 4. Methodology and Work Plan information access for each type of cognitive disability. Module B. Text summarization system: COMThe research work that would be addressed in the PENDIUM CLEAR.TEXT project will have a duration of two years, Based on the expertise and previous research carried starting from 1st September 2022. out by the team of the project, together with the advancements of the state of the art in this task -[? ], [? ], [? ]- the non-literal language. main goal of this module is to adapt the COMPENDIUM Task D.1. Coreference resolution. One important protool [? ] to summarize administrative texts written in cess in the conversion of documents into a form facilitatSpanish. Therefore, the following two tasks are included ing reading comprehension is coreference resolution, in in this module. which links between pronouns and noun phrases refer
Task B.1. Adapting COMPENDIUM to summarize docu- ring to the same concept are automatically identified. ments of the public administrations. The main objective Task D.2. Word sense disambiguation. One prevalent of this task is to adapt and fine-tune COMPENDIUM’s source of ambiguity in language is word sense ambiguity, approach to enable efective summarization of Spanish that is, when a single word may have any one of several text documents generated by public sector organizations. meanings depending on the context in which it is used. The type of summary to be generated will be abstractive, Task D.3. Processing of dificult words . The language which means that not only the relevant information will used in public sector documents tends to be that of a be extracted, but also, such relevant information will be sophisticated and technical nature. These documents paraphrased using diferent vocabulary and structures, usually contain archaisms, highly technical language, guaranteeing that the same meaning in the generated acronyms or aphorisms. The objective addressed by this text. task, via constructing or adapting a complex word pro
Task B2. Integrate COMPENDIUM in an accessible cessing tool, will be the detection and evaluation of the and easy-to-use platform. Once the COMPENDIUM tool degree of complexity of these terms and their substitution has been adapted for Spanish public sector documents by other equivalent words that are more easily under(task B.1), the goal of this task is to integrate the gen- stood. erated software into a user-friendly platform, so that Module E. Generation of simplex Documents it can be accessible and easy to use. For this, we will This project will be based on converting documents take into account the Web Accessibility Guidelines de- into an accessible personalised form for end-user profiles ifned by the World Wide Web Consortium (W3C) 3, not with cognitive dificulties. only for the design of the Web interface, but also for Task E.1. Language processing components. One aspect the best way to output the content generated as the of this functionality will be achieved by the implementaresulting summary, meeting the standards defined in: tion of language processing components aimed at removhttps://www.w3.org/WAI/standards-guidelines/wcag/. ing obstacles to reading comprehension while preserving Module C. Processing of structural complexity the meanings expressed by the original document. The This module aims at to analyse and investigate the language processing components implemented in the state of the art of three main areas of natural language project will detect and remove obstacles caused by strucprocessing, and their application to the simplification tural complexity (Module B) and ambiguity in meaning of texts. These areas are lexical processing, syntactic (Module C). processing, and discourse processing. Task E.2. Generation of simplex documents. The second
Task C.1. Lexical processing. The detection and removal aspect of the conversion process will be the generation of of obstacles to reading comprehension caused by the content consist in textual data occurring in the document individual words used in documents that users wish to to supplement that textual data. This content will include access will build on a variety of related research. An document navigation tools, summaries, pre-reading quesexhaustive review of the state of the art will be carried tions, and images. out to determine the best lexical tools. Task E.3. SIMPLE.TEXT System. Construction of a sys
Task C.2. Syntactic processing. Many technologies to tem that integrates the set of tools developed into a platdetect and remove syntactic obstacles to reading compre- form that is accessible and easy to use. hension depend on syntactic parsing. Syntactic relations Module F. Evaluation and quality such as coordination and subordination, signaled by us- Task F.1. Evaluation and quality. The objective of this ing conjunctions and punctuation, must be recognised task is to ensure the quality of the technology and results accurately. An exhaustive review of the state of the art obtained in previous tasks. Both intrinsic and extrinsic will be carried out to determine the best syntactic tools. evaluation methods will be performed. The implemented Module D. Processing of ambiguity in meaning system (SIMPLE.TEXT system) will be evaluated with Various lines of research in the state of the art of respect to the accessibility of the personalised documents processing ambiguity in meaning are relevant to the that it generates by means of reading comprehension testproject. These include related work in three main areas: ing. The usability of the system by people with cognitive i)anaphora and coreference resolution, ii) the process- disabilities will be evaluated, and a qualitative research ing of word sense ambiguity, and iii) the processing of methodology employed to assess the contribution to inclusion resulting from the use of the technology. 3https://www.w3.org/WAI/fundamentals/accessibility-intro/ Module G. Project coordination
of internal communication, evaluate progress and, if necessary, readjust the objectives. To successfully achieve this goal, one task is proposed (Task G.1. Project monitoring).
Module H. Project dissemination
This module is concerned with coordinating and supervising the dissemination of the project through publications and other means.
Task H.1. Dissemination of the project. The dissemination of the project, as well as the derived results, will be carried out during all stages of the project. The plan for communication, dissemination and exploitation of results will be articulated on three levels: i) information on the project and its development through a dedicated website; ii) scientific publications with research results, and iii) dissemination and exploitation of resources and prototypes created in the research project.
Language technologies are at the cornerstone of AI and are among those tools for which there will be the greatest demand in the next decade. Concerning the scientific and technical impact, our project focuses on research and development of technologies on NLP to support the authoring of accessible Spanish content for public sector organisations that is intelligible to people with cognitive disability. Applications derived from this research include the following: texts summarisation; texts simplification; lexical and syntactic analysis; anaphora resolution; word sense disambiguation; the creation of summarization reports providing explanations; and facilitating the comprehension of complex expressions (e.g. metaphors) in public sector organizations domains. The methods, models, resources, and systems that will be researched, developed and deployed in the project will attract the interest of the NLP and AI research communities.
We can identify the following positive social and economic impacts for people with cognitive disabilities that would result from being granted the opportunity to develop the proposed project: • Social impact for those with cognitive disabilities: – Promotes a cooperation between the technological and social fields, with a view to the design of technological solutions that consider the needs of people with disabilities. – Facilitates the daily actions of people with
disabilities. – Widens inclusion and empowerment in Eu
rope. – Improves the quality of life of those with cognitive disabilities, enhancing their access to educational, vocational, cultural, and social opportunities in Europe. – Promotes an independent life and the ca
pability to realize personal goals. – Enables access to a meaningful education. – Encourages personal involvement in all de
cisions that afect their future. – Promotes participation in the benefits offered by cultural, recreational and sporting activities. • Economical impact for those with cognitive disabilities: – Facilitates accessibility to digital information to promote economic and political inclusion. – Promotes full labor inclusion by 2024 through increased employment opportunities for individuals with cognitive disabilities and improved productivity by facilitating their ability to perform work-related tasks. – Widens inclusion and empowerment in Eu
rope. – Enables participation in the services provided for promoting efective management of personal finances. – Provides equal access to and use of all facilities, services and activities in the public. sector organizations at local, regional and national level, such as filing tax returns, payment of fines, community charges, etc.