A shortcoming has been found in search as revealed by Google Research paper

A new study sheds light on a little-known aspect of information retrieval that Google has not yet mastered.

It has been showed by Google Research paper on Long Form Question Answering that how tough it is to respond to questions that require more thorough and detailed responses. In this particular area, the researchers were able to enhance this type of question answering, they also acknowledged that their findings could be much better.

This study focuses on Long Form Open-Domain Question Answering, a field in which Natural Language Processing is still improving. Factoid is a term for what search engines excel at. Factoid open-domain Question answering or simply Question answering in the open-domain.

Open Domain Query Answering is a process in which an algorithm provides a natural response to a question.

Long-form Question Answering (LFQA) is necessary but difficult, according to the research paper, and progress in achieving this type of question answering is just not as far along as Open-domain Question Answering.

As per research paper, Long-form question answering (LFQA) in the open domain is a crucial task in natural language processing (NLP) that entails retrieving supporting information and utilizing them to produce an extensive lengthy response. Although, there has been significant notable development in factoid open-domain question answering (QA), where a single short term or entity makes good sense to answer a question, there has been considerably limited progress in the case of long-form question answering. LFQA is still an essential activity as it serves as a test platform for determining the validity of generative text templates.

On Google search engine, a searcher usually asks a question, and the search engine responds with a fairly short chunk of information. A simple question like asking for contact number, is what search engine excel in, particularly because the response is logical rather than descriptive.

Long-form question answering is more difficult since the questions require paragraph-length responses rather than brief sentences. Facebook is now also focusing on long-form question answering and has provided with some interesting approaches, such as using the Explain Like I'm 5 question and response (a dataset called ELI5). Facebook also acknowledges that there is still work to be done.

Such kinds of long-form queries are hard for Google (and Bing) to address. This might have had an effect on their potential to produce material that highlights difficult questions.



Read next: Google Earth Teams Up with NASA and Others to Form a Timelapse Feature Showing How Earth Is Being Affected by Climate Change
Previous Post Next Post