Dealing with decisional natural language why-question in business intelligence

Guessoum, Meriem Amel; Djiroun, Rahma; Boukhalfa, Kamel

doi:10.1109/iacs.2017.7921945

Cited by 4 publications

(13 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…"Why has the internet sales amount decreased?". This question type is interesting, allowing decision-makers to understand some decision-making indicators such as the cause or origin of a trend, causes of customers' behavior [16]. Indeed, a Why-Question allows decision-makers to perform diagnostic analysis on the data of the Warehouse.…”

Section: Introductionmentioning

confidence: 99%

“…These concepts are crucial in a decision-making system. To address a NL Why-Question emitted in BI context, we have proposed in [16] an approach that aims at providing decision-makers with answers that allow them to detect factors that influence a phenomenon, for effective decision support. Our approach emphasizes on both the decision maker's requirement and the DW (concepts and content).…”

Section: Introductionmentioning

confidence: 99%

“…It is based on a mathematical model that performs a deep analysis on the Warehouse's data. Since the answering process is closely related to the decision maker's requirement, we have proposed in [16] a Why-Question's model that captures the necessary components and constraints (a Why-Question must comport a measure, trends, temporal or not temporal dimensions) upon which our approach is built. In our approach [16], the decision-maker formulates freely, his Why-Question through using his own terms in NL.…”

Section: Introductionmentioning

confidence: 99%

“…(2) -When the Why-Question doesn't conform to the model that defines it, the approach [16] does not generate answers. For example: "Why didn't the company evolve?…”

Section: Introductionmentioning

confidence: 99%

“…This grammar is a pattern that aims at formalizing the model that specifies the desired content of a Why-Question. Indeed, our grammar captures all the restrictions defined in the Why-Question's model [16]. To build this grammar, we adopt first the linguistic patterns proposed by "BARGUI and al" in [4,5] and then we extend them with respect to our Why-Question's model.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Natural language why-question in Business Intelligence applications: model and recommendation approach

et al. 2022

View full text Add to dashboard Cite

Business Intelligence is the key technology for users to effectively extract valuable information from oceans of data for decision-making. Data warehouses and on-line analytical processing systems have therefore been developed to contribute effectively to the decision-making process. To extract information that is useful to decision-making, decision-makers express their needs in natural language. Such requirements may be formulated in natural language interfaces in free syntax, avoiding unfamiliar language (SQL, MDX). Natural Language queries can stand for WH-questions ("What, Who, Where, Why, etc.") or a set of Keywords. In this paper, we emphasize on the "Why-Question". This type of question provides answers that help in the diagnosis analysis of the data Warehouse. To deal with a Why-Question, we propose a model that mainly captures the components that reflect the multidimensional aspect of the Data Warehouse. When a decision-maker formulates his Why-Question in natural language, he uses his own terms. This decisional need can be not precise because the decision-maker is not always aware of the Data Warehouse's lexicon as well as the Why-Question's model. Consequently, the decision-maker must reformulate his initial question. Otherwise, the Why-Question's answering process will not be triggered. This situation is not obvious for a decision-maker, especially when the reformulation of the question becomes iterative. To handle these issues, we lean towards a Why-Question's recommendation approach based on both the Data Warehouse's content and the decision maker's requirement. This proposal aims to recommend to the decision-maker a set of natural language Why-Questions instead of rephrasing his initial question. To guide the recommendation process, we rely on a grammar that formalizes the Why-Question's model. To validate our approach, a tool called "WQ-Recommender" is developed. An experimental study is presented to evaluate the relevance of the proposal.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…(2) -When the Why-Question doesn't conform to the model that defines it, the approach [16] does not generate answers. For example: "Why didn't the company evolve?…”

Section: Introductionmentioning

confidence: 99%