ES2847588A1 - System to improve the user experience without abstraction capabilities in the information query (Machine-translation by Google Translate, not legally binding) - Google Patents

Abstract

Procedure and system to provide help to users who do not have the ability to abstract concepts when they need to search for information. To solve this problem, a system is proposed that dialogues with the user to extract knowledge about aspects of the information that attract their attention and the feeling and emotion that they arouse, all through natural language, relying on voice-text and text conversion techniques. -voice. A conversational assistant or chatbot with the ability to generate natural language leads the dialogue about the information. Using artificial intelligence techniques, the significant words of the user's expressions are extracted and their feelings and emotions are analyzed, and new searches are carried out by combining the significant words of the user's expressions with the significant words contained in the news, if applicable. that the feeling and emotion elicited in the user are positive. (Machine-translation by Google Translate, not legally binding)

Description

DESCRIPTION

System to improve the user experience without abstraction capabilities in the information query

Technical sector

The present invention is framed in the field of conversational assistants capable of generating, by means of artificial intelligence, dialogues in natural language.

Background of the invention

Recent advances in artificial intelligence have allowed chatbots to automatically generate responses to user requests [1]. As a result, chatbots are currently used in business [2], education [3], and medicine [4]. However, mimicking human intelligence is still out of reach. Consequently, chatbots are used for very specialized tasks and their interactions are targeted, precise, and brief. In addition, users find it easier to communicate with a chatbot if it uses short messages, and thus they remain interested longer [5].

The domain of the present invention is to provide entertainment and information to users who do not have the ability to abstract concepts when they need to search for information.

Some entertainment chatbots, like EduRobot [6], sing and tell stories. In the health field, there are online platforms for general medical care such as HealthTap [7], Oscar [8] and Molly [9], through which doctors and patients interact. None of these systems solves the inability to abstract information of some groups of users. In fact, for older people, the focus is often on medical care rather than entertainment. Although there are news reporter chatbots [10], they have not been endowed with empathy or skills to deal with people with difficulties in abstracting information.

With regard to the prior art protected by patents, in [11] it is proposed to increase the search expressions using machine learning techniques, but the user is not encouraged to enter information through an artificial dialogue, nor are their feelings analyzed through artificial intelligence techniques. In [12] a system is proposed to facilitate online navigation for the elderly through voice commands, but, in case of problems (such as concept abstraction problems), the system relies on human volunteers who access line. In [13] a terminal with a graphical interface is proposed in which an avatar guides a user through the application menus by voice, including those of information search engines, but the invention does not have intelligent conceptual abstraction capabilities. In [14] a system is proposed to provide information to users who have Internet access problems.

It uses an interface based on voice-text conversion to receive user requests, which are transformed into online information inquiries. This system estimates the degree of interest of the users, although instead of natural language processing, the inflections of the voice, facial expressions and oral expression are analyzed. In addition, this system uses an avatar-type graphical interface that adapts to the perceived feeling, as well as a conversational assistant to present the information to the user. However, in this invention the conversational system is not used to guide the user (dialogue with him / her to deduce his / her interests in order to carry out searches), but rather searches are carried out in based on information collected from the user before starting the conversation. The wizard is used to present the information to the user and generate a profile of the user for future searches. That is, the system does not have automatic conceptual abstraction capabilities. In [15] a system is proposed that collects personal information from the user and, later, when the user searches for information, adjusts them based on the personal information collected. That is, the search settings are personalized with user information collected a priori.

References:

[1] Xu, A., Liu, Z., Guo, Y., Sinha, V., & Akkiraju, R. (2017). A new chatbot for customer Service on social media. In Proc. Conference on Human Factors in Computing Systems, pp. 3506 3510, 2017.

[2] Thomas, N. T. (2016). An e-business chatbot using AIML and LSA. In Proc. International Conference on Advances in Computing, Communications and Informatics, pp. 2740-2742.

[3] Robins, B., Dautenhahn, K., Te Boekhorst, R., & Billard, A. (2005). Robotic assistants in therapy and education of children with autism: can a small humanoid robot help encourage social interaction skills ?. Universal access in the information society, 4 (2), pp. 105-120.

[4] Kumar, V., & Keerthana, A. (2016). Sanative Chatbot For Health Seekers. International Journal Of Engineering And Computer Science, 5, pp. 16022-16025.

[5] Hill, J., Randolph Ford, W., & Farreras, I. G. (2015). Real conversations with artificial intelligence: A comparison between human human online conversations and human chatbot conversations. Computers in Human Behavior 49, pp., 245-250.

[6] Budiharto, W., Cahyani, A. D., Rumondor, P. C., & Suhartono, D. (2017). EduRobot: Intelligent Humanoid Robot with Natural Interaction for Education and Entertainment. Procedia Computer Science 116, pp. 564-570.

[7] https://www.healthtap.com

[8] https://www.hioscar.com/

[9] http://www.senselv.com/

[10] Matsumoto, R., Nakayama, H., Harada, T., & Kuniyoshi, Y. (2007). Journalist robot: Robot system making news articles from real world. In 2007 IEEE / RSJ International Conference on Intelligent Robots and Systems, pp. 1234-1241. IEEE.

[11] Alternative search query prediction. Patent CN101268463A, China.

[12] One-stop type intelligent comprehensive knowledge Service system for the elderly based on cloud computing. Patent CN103914794A, China.

[13] The ¡Information terminal device. Patent JP2004013836A, Japan.

[14] Electronic personal interactive device that determines topics of interest using a conversational agent. Patent US9634855B2, United States.

[15] Method and device for improving accessability to information displayed on a Computer. Patent US20110072257A1, United States.

[16] Su, M. H., Wu, C. H., Huang, K. Y., Hong, Q. B., & Wang, H. M. (2018). A chatbot using LSTM-based multi-layer embedding for elderly care. In Proceedings of the 2017 International Conference on Orange Technologies, ICOT 2017, pp. 70-74.

[17] Fernández Gavilanes, M., Álvarez López, T., Juncal Martínez, J., Costa Montenegro, E., González Castaño, F. J. Unsupervised method for sentiment analysis in online texts. Expert Systems with Applications 58 (2016), pp. 57-75.

[18] García Méndez, S., Fernández Gavilanes, M., Costa Montenegro, E., Juncal Martínez, J., González Castaño, F. J. (2019). A library for automatic natural language generation of Spanish texts. Expert Systems with Applications 120, pp. 372-386.

Explanation of the invention

The present invention is concerned with providing help to users who do not have the ability to abstract concepts when they need to search for information. Current information search systems on the Internet assume that the user has this capacity, but some groups (very old people, people with cognitive diversity) lack it. In addition, for these groups it can be difficult to search for information through textual interfaces or even to consult the information in a textual way. To solve this problem, a system is proposed that dialogues with the user to extract knowledge about information presented in the form of news that attracts their attention and the feeling and emotion that it arouses, all through natural language, relying on voice-to-conversion techniques. text and text-voice.

Thus, the present invention uses a novel method and system that will improve the user experience when consulting information from online sources, especially suitable for users who experience difficulties in abstracting the concepts necessary to perform searches.

The central element is the conversational assistant or chatbot that induces the user to express their interests and opinions through simple messages, expressed in a natural voice. Depending on the phase of the dialogue (determined by the chatbot), the opinion behind the user's expressions in natural voice, analyzed using artificial intelligence techniques, and the keywords contained in said expressions, an interest analyzer directs a information search engine on the Internet. The interest analyzer makes its decisions based on the reactions produced by the news in the user, detected by artificial intelligence techniques.

Therefore, using artificial intelligence techniques, the significant words of the user's expressions are extracted and their feelings and emotions are analyzed, and new searches are carried out by combining the significant words of the user's expressions with the significant words contained in the news. , in case the feeling and emotion elicited in the user are positive. That is, if we compare our proposal with the state of the art, the dialogue conducted by the system itself is used to deduce the interests of the user in order to carry out searches, instead of them depending on information collected from the user before starting the conversation .

Preferred embodiment of the invention

The description and drawings illustrate the principles of the invention. Although the present invention has been described with reference to specific embodiments, those skilled in the art should understand that both such modifications and their details may be subject to change without departing from the scope of the present invention as defined by the claims.

Furthermore, all the examples mentioned herein are primarily and expressly intended for educational purposes to help the reader understand the principles of the invention and the concepts contributed by the inventor (s) to further the art, and should be construed without limitation to the examples and conditions specifically mentioned. Furthermore, all statements herein that mention principles, aspects, and embodiments of the invention, as well as their specific examples, are intended to encompass their equivalents.

The present method and system refer to an electronic device of any type, such as a mobile phone, a laptop, a desktop computer or a device manufactured on purpose, with access to a graphic screen (101), a microphone (102), a loudspeaker (103) and on-line information sources, such as the Internet (104), which executes a conversational assistant or intelligent chatbot capable of holding dialogues with a user to help them abstract concepts in order to carry out information searches. To do this, the intelligent assistant interacts with a series of functional elements that provide artificial intelligence and natural language processing functions.In some embodiments, one or more of the aforementioned hardware peripherals will not be included in the device itself, but will be external peripherals. of the same, or the device will connect with other devices that have said peripherals to use them when needed.

In one embodiment, Figure 1 shows the elements of the invention. On the screen (101) a symbolic representation of a character is shown that reflects moods and emotions (105). In the example of the realization, the character reflects sadness. The screen (101) shows a text that the device reproduces as a synthetic voice through the loudspeaker (103). A shift indication icon (107) with two states is shown on the screen. One of the states tells the user when he can speak. The other indicates that it cannot.

In one embodiment, Figure 2 shows the functional blocks involved in the smart functions of the device.

The conversational assistant or chatbot (205) induces the user to express their interests and opinions through simple messages (201), which are transformed into voice by means of a text-voice converter (202). In one embodiment, the chatbot dialogs can be implemented using AIML language.

User responses are translated into text by a voice-text converter (203). Depending on the phase of the dialogue, determined by the chatbot, and the opinion behind the user's expression (detected by means of a sentiment and emotion analyzer (204)), the system decides whether the chatbot (205) produces the following response (201) directly based on the application of rules or in real time through a natural language generator (206). In one embodiment, artificial intelligence techniques can be used [16]. The natural language generator uses as inputs the keywords contained in the user's response (extracted by the keyword extractor (207)) and the output of the sentiment analyzer (204). In one embodiment, the keyword extractor removes words with little or no semantic load (determiners, prepositions, conjunctions, etc., as well as general-purpose pronouns and verbs, for example), removes redundancies through stemming, and the remaining words are the keywords. The state of the art field that defines this type of technique is natural language processing. In one embodiment, the feeling and emotion analyzer can use techniques such as those described in [17] and its references. In one embodiment, the chatbot can rely on an external natural language expander such as the one described in [18] to generate its responses.

In order to make an automatic abstraction of the user's interests to offer him information, an interest analyzer (208) decides the searches of a search engine (209) for information on the Internet. The interest analyzer makes its decisions based on the reactions produced by the news in the user (consulting the output of the sentiment and emotion analyzer (204)) and the semantic proximity of the content of the user's expressions (provided by the extractor of keywords (207)) to the news previously presented to the user. A news extractor (210) extracts the news from Internet sites identified by the search engine (209). In one embodiment, the news extractor is a scrapper that extracts text from newspaper, portal, blog or newscast web pages. In one embodiment, the interest analyzer (208) can select the user's words that belong to the same semantic category as the main words of the news (excluding words with no or low semantic load) by means of the Multilingual Central Repository tool.

In one embodiment, the device conducts the dialogue by displaying a green icon on the screen when a response is expected from the user and a red icon when the device itself is speaking (107).

In one embodiment, when the device is speaking it displays the transcript (106) of its speech.

In one embodiment, for a certain feeling or certain emotions of the user or the utterance of the device, the avatar displays a corresponding expression (105) using the usual graphic sentiment coding conventions, such as emoticons.

Claims (12)

1. A procedure to improve the user experience when consulting information on the Internet through automatic abstraction capabilities, comprising said procedure: a) Reproduction by synthetic voice of information extracted from the Internet; b) Voice dialogue with the user about said information through a chatbot that induces the user to express what aspects of the information attract his or her attention and what opinions are raised by said aspects; c) Extraction of significant words from the user's expressions; d) Extraction of the feeling and emotion of the user's expressions; e) Search for new Internet content based on the relationship between the significant words contained in the user's expressions and the news reproduced, on the one hand; and the output of an analyzer of the feeling and emotion aroused in the user, by another; enhancing the relevance of words when said feelings and emotions are positive. 2. A device executing the method of claim 1. 3. The device according to claim 2, wherein the device communicates with the user through speech conversion and synthesis. The device according to claim 2, wherein the screen of the device presents an avatar whose expression varies according to the feeling and emotion of the dialogue. 5. The device according to claim 2, wherein the chatbot has the ability to generate natural language based on the user's previous expressions, their feelings or their emotions. 6. The device according to claim 2 where an Internet service specialized in searches such as Google performs information searches. The device according to claim 2 wherein the device extracts information from Internet sites by means of a scrapper. 8. The device according to claim 2 wherein the device obtains information directly from Internet text services. The device according to claim 2, wherein the device transcribes multimedia information from Internet services or sites into text. The device according to claims 1-9, wherein the device is a mobile phone, a laptop, a portable computer, a tablet, a desktop PC or a television. The device according to claims 1-10, wherein the device has a screen, a speaker and a microphone. 12. The device according to claims 1-11, wherein the device uses the screen, the speaker, the microphone or the Internet connection of other nearby devices.