Charting the Evolution and Future of Conversational Agents: A Research Agenda Along Five Waves and New Frontiers

Classified as probably not a substantial claim. May combine two claims.

The paper provides an overview of leading authors and countries and demonstrates the development of keywords in the CA research domain.

“In our study, we provide an overview of leading authors and countries and demonstrate the development of keywords in this research domain.”

Classified as probably not a substantial claim.

The paper's results highlight fundamental technical evolutions and theoretical paradigms in CA research.

“Our results highlight fundamental technical evolutions and theoretical paradigms in CA research.”

May combine two claims.

The majority of the identified literature reviews focus on studies from 1998 onward, thereby neglecting first papers on CAs such as ELIZA.

“The majority of reviews focus on studies from 1998 onward, thereby neglecting first papers on CAs such as well-known first text-based agent ELIZA (Weizenbaum, 1966).”

• cites Weizenbaum, 1966 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The purple-colored research cluster is defined by authors such as Nass and is concerned with the personality of CAs.

“The purple-colored research cluster is defined by authors, such as Nass, and is concerned with the personality of CAs.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Technical models, such as NLP, deep learning, and machine learning, appear in the chatbot cluster.

“Technical models, such as NLP, deep learning, and machine learning, appear here.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Approximately 30% of the 50 most prominent keywords illustrate the research focus.

“Approximately 30% of the 50 most prominent keywords illustrate the research focus.”

May combine two claims.

E-learning and mental health represent relevant interaction contexts, which have also seen great prominence in recent publications.

“E-learning and mental health represent relevant interaction contexts, which have also seen great prominence in recent publications (Androutsopoulos et al., 1995; Chaves & Gerosa, 2021; Følstad & Skjuve, 2019; Hauswald et al., 2016).”

• cites Androutsopoulos et al., 1995 sourceNot checked yet

  Not checked in the free run.

• cites Chaves & Gerosa, 2021 sourceNot checked yet

  Not checked in the free run.

• cites Følstad & Skjuve, 2019Not checked yet

  Not checked in the free run.

May combine two claims.

The most prominent keywords are text-based CAs, with chatbot(s) being most frequent, followed by embodied CAs.

“The most prominent are text-based CAs, with the chatbot(s) being most frequent, followed by embodied CAs.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The co-cited reference network is defined by four key clusters.

“Our co-cited reference network, defined by four key clusters, allows us to better understand how the field was built and which theoretical foundations mainly drove CA research throughout time.”

Probably not a substantial claim: may not pass S3 (load-bearing).

With 91 articles, Pelachaud is the most productive author.

“With 91 articles, Pelachaud is the most productive author.”

May combine two claims.

All other keywords, such as Human-Computer Interaction or NLP, have been relevant over the last couple of years on a continuous level and to a high frequency.

“All other keywords, such as Human-Computer Interaction or NLP, have been relevant over the last couple of years on a continuous level and to a high frequency.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The term 'voice user interface' appears within the CA cluster, yet is not linked to any further words.

“The term “voice user interface” appears within the CA cluster, yet is not linked to any further words.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Smart personal assistants, such as Amazon's Alexa, are linked to the term 'conversational agent'.

“Lastly, smart personal assistants, such as Amazons Alexa, are linked to the term as well.”

May combine two claims.

In general, authors from the same country published together, with an average collaboration score of 2.46.

“In general, authors from the same country published together, with an average collaboration score of 2.46.”

May combine two claims.

Germany has 2,595 total citations and is the country with the second-most citations.

“A cluster of countries follows with a smaller number of total citations, with Germany with 2,595 total citations as the country with the second-most citations.”

Probably not a substantial claim: may not pass S3 (load-bearing).

12 out of the 50 top keywords indicate the interaction context of a publication.

“Lastly, keywords indicate the interaction context of a publication (12 out of the 50 top keywords).”

Probably not a substantial claim: may not pass S3 (load-bearing).

The number of studies analyzed in the identified literature reviews ranges from 29 articles to 262 articles.

“The number of studies analyzed in literature reviews ranges from 29 articles (Rheu et al., 2021) to 262 articles (Diederich et al., 2022).”

• cites Rheu et al., 2021 sourceNot checked yet

  Not checked in the free run.

• cites Diederich et al., 2022Not checked yet

  Not checked in the free run.

May combine two claims.

Approximately 26% of the top keywords provide insights into the statistical models, pointing toward the importance of technical advances for the proliferation of CAs.

“Approximately 26% of the top keywords provide insights into the statistical models, pointing toward the importance of technical advances for the proliferation of CAs.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The third most productive author is Griol, with 47 published articles.

“The third most productive author is Griol, with 47 published articles.”

May combine two claims.

Deep learning has been relevant from the beginning onward, similar to the keyword dialogue system.

“Lastly, deep learning has been relevant from the beginning onward, similar to the keyword dialogue system.”

Probably not a substantial claim: may not pass S3 (load-bearing).

AI, machine learning, and natural language (processing) dominate the keywords.

“AI, machine learning, and natural language (processing) dominate the keywords, which also illustrates how this technical interplay makes the architecture of CAs unique.”

Probably not a substantial claim: may not pass S3 (load-bearing).

An equally extensive keyword cluster exists around the term 'chatbot'.

“An equally extensive keyword cluster exists around the term “chatbot”.”

May combine two claims.

The keywords affective computing, embodied CA, emotion, natural language interface, and virtual agent have no linear or clear development, but all increase over time.

“This is different from the keywords affective computing, embodied CA, emotion, natural language interface, and virtual agent.These keywords have no linear or clear development, but all of them increase over time.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Tasks such as question answering and information retrieval are associated with CAs.

“Tasks such as question answering and information retrieval are associated with CAs.”

Probably not a substantial claim: may not pass S3 (load-bearing).

In the context of embodied conversational agents, the personality and emotions of agents are commonly discussed.

“In this context, the personality and emotions of agents are commonly discussed.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The term 'embodied conversational agent' is more concerned with additional modalities, including the avatar or facial expression of the agent.

“The term “embodied conversational agent”, is more concerned with additional modalities, including the avatar or facial expression of the agent.”

May combine two claims.

The most relevant research focuses are trust and anthropomorphism, and agent personality and emotions.

“The most relevant research focuses are trust and anthropomorphism, and agent personality and emotions.”

May combine two claims.

With 61 published articles, Bickmore is the second most productive author, publishing between 2002 and 2020.

“With 61 published articles, Pelachaud is followed by Bickmore as the second most productive author who published between 2002 and 2020.”

May combine two claims.

Whereas Eastern players (i.e., Baidu, ASAPP) have increased in appearance, western players still dominate.

“Whereas Eastern players (i.e., Baidu, ASAPP) have increased in appearance, western players still dominate.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The development of certain keywords, particularly AI, machine learning, natural language processing, neural network, and deep learning, showed either a huge spike in the last five years or continuous steep development.

“We can observe this trend quite clearly when looking at the development of certain keywords, particularly”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Traditional text-based conversational agent applications have started to shift to voice-based conversational agents.

“Moreover, traditional text-based CA applications have started to shift to voice-based CAs.”

May combine two claims.

Publications by corporate affiliations were dominated by IBM and Microsoft until 2008, followed by Google from 2009 until 2011.

“Publications by corporate affiliations were dominated by IBM and Microsoft until 2008, followed by Google from 2009 until 2011.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

The term 'conversational agent' does not necessarily seem to act as an overarching term for the different types of agents.

“The term “conversational agent” does not necessarily seem to act as an overarching term for the different types of agents.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

The keywords AI, chatbot, and human-computer interaction started to emerge in 2000-2005.

“For instance, the keywords”

May combine two claims.

Concluding the fourth wave, the authors observed a stronger focus on natural language understanding and the emergence of novel and advanced voice-based conversational agents.

“Concluding the fourth wave of CA research, we have observed a stronger focus on natural language understanding and the emergence of novel and advanced voice-based CAs.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Certain tasks and contexts are connoted with certain terminology.

“Moreover, certain tasks and contexts are connoted with certain terminology.”

May combine two claims.

From 2012 onwards, publications by commercially affiliated researchers have become increasingly fragmented through a greater number of players and Asian companies such as Baidu becoming apparent.

“From 2012 onwards, publications by commercially affiliated researchers have become increasingly fragmented (i.e., through a greater number of players and Asian companies, such as Baidu, becoming apparent).”

May combine two claims.

There has been a steady increase in the frequency of corporate-related keywords since 2000 and a steep increase since 2017.

“Turning toward the temporal development of corporate-related keywords, we identified a steady increase in the frequency of these keywords since 2000 and a steep increase since 2017 (see Fig. 4).”

May combine two claims.

Google is most frequently listed as a keyword in 2020 (N = 78), followed by Metfoot_0 (N = 49) and Microsoft (N = 41).

“More specifically, Google is most frequently listed as a keyword in 2020 (N = 78), followed by Metfoot_0 (N = 49) and Microsoft (N = 41).”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Initial research driven by the CASA paradigm and the Turing test found great prevalence in contemporary studies.

“In a similar vein, initial research driven by the CASA paradigm and the Turing test found great prevalence in contemporary studies.”

May combine two claims.

Only one comparable study, Io and Lee (2017), analyzed approximately 4200 articles, but only from 1998 onwards.

“We find only one comparable study from Io and Lee (2017), who analyzed approximately 4200 articles, however only from the year 1998 onwards.”

• cites Io and Lee (2017)Not checked yet

  Not checked in the free run.

May combine two claims.

While research into voice-based conversational agents has increased over the past few years, it has not yet been established as research around chatbots.

“While research into voice-based CAs has increased over the past few years, it has not yet been established as research around chatbots.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The bibliometric analysis corpus draws from 1,920 sources.

“Sources (Journals, Books, etc.) | 1,920”

Probably not a substantial claim: may not pass S3 (load-bearing).

The corpus includes 3,686 conference papers.

“Conference papers | 3,686”

Probably not a substantial claim: may not pass S3 (load-bearing).

Certain tasks, contexts, and technical models are commonly associated with different terms.

“Interestingly, certain tasks, contexts, and technical models are commonly associated with different terms.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The conversational agent does not necessarily seem to act as an overarching term for the different types of agents; certain tasks and contexts are connoted with certain terminology.

“The CA does not necessarily seem to act as an overarching term for the different types of agents.Certain tasks and contexts are connoted with certain terminology.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The average citation per document is 8.047.

“Average citation per document | 8.047”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

A voice-based conversational agent finds little appearance in the keyword cluster, which may be the result of the choice of dominant keywords or voice-based agents not having been researched as extensively as text-based ones.

“A voice-based CA finds little appearance in our keyword cluster, which may be the result of a certain choice of dominant keywords or voice-based agents not having been researched as extensively as text-based ones.”

May combine two claims.

Most related studies are focused on literature only from the year 2000 or newer and do not include more than 100 articles.

“Looking at related studies (see Appendix B), we find that most research is focused on literature only from the year 2000 or newer and does not include more than 100 articles.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The corpus includes 130 book chapters.

“Book chapters | 130”

May combine two claims.

All other comparable studies have a more traditional literature review approach and seem to focus on some specific topic.

“All other studies we find to be comparable have a more traditional literature review approach and seem to focus on some specific topic.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The average citation per year per document is 1.237.

“Average citation per year per document | 1.237”

Probably not a substantial claim: may not pass S3 (load-bearing).

The corpus includes 1,290 articles.

“Articles | 1,290”

Probably not a substantial claim: may not pass S3 (load-bearing).

The average years from publication for the documents is 5.81.

“Average years from publication | 5.81”

Probably not a substantial claim: may not pass S3 (load-bearing).

The bibliometric analysis corpus comprises 5,106 documents.

“Documents | 5,106”

Probably not a substantial claim: may not pass S3 (load-bearing).

The corpus has 14.801 Keywords plus (ID).

“Keywords plus (ID) | 14.801”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Most current CAs do not include personalized features, trust or explainable AI features, or believable humanization.

“Most current CAs do not include personalized features, trust or explainable AI features or feature believeable humanization”

Probably not a substantial claim: may not pass S3 (load-bearing).

Bickmore and Picard (2005) has 581 citations.

“Bickmore and Picard (2005) | 581”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Most CA software, programs, libraries, and models are free and open source.

“Most CA software, programs, libraries, models, etc. are free and open source.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

Many companies use CAs, but most are simple clickbots or less advanced CAs that may even hurt the interaction.

“Many companies use CAs but most being simple clickbots or less advances CAs that may even hurt the interaction.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The corpus has 8.327 authors' keywords (DE).

“Authors' keywords (DE) | 8.327”

Probably not a substantial claim: may not pass S3 (load-bearing).

The first chatbot was developed in 1966 by Weizenbaum.

“Historically, the first chatbot was developed in 1966 by Weizenbaum (1966).”

• cites Weizenbaum (1966) sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

With the advent of AI and the deployment of more complex models such as neural networks, CAs offer the potential to be sophisticated interactive systems, assisting customers to handle complaints and acting as learning assistants in tutoring students or employees.

“With the advent of artificial intelligence (AI) and the deployment of more complex models such as neural networks (Sedik et al., 2022;Goudos et al., 2019;Kushwaha & Kar, 2021), CAs offer the potential to be sophisticated interactive systems, assisting not only customers to handle their complaints but also acting as learning assistants in tutoring students or employees, for instance (Schmitt et al.”

• cites Sedik et al., 2022Not checked yet

  Not checked in the free run.

• cites Goudos et al., 2019 sourceNot checked yet

  Not checked in the free run.

• cites Kushwaha & Kar, 2021Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Research on CAs is a rapidly growing field, with an increase in publications over the last few decades, especially fueled through the development of CAs such as Amazon's Alexa that build upon voice as a modality.

“Research on such agents is a rapidly growing field, with an increase in publications over the last few decades, especially fueled through the development of CAs such as Amazon's Alexa that build upon voice as a modality, so called voice bots (Schmitt et al., 2021;Seaborn et al., 2022;Kendall et al., 2020).”

• cites Schmitt et al., 2021Not checked yet

  Not checked in the free run.

• cites Seaborn et al., 2022 sourceNot checked yet

  Not checked in the free run.

• cites Kendall et al., 2020 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

CAs' adoption can be impeded by communication and interaction problems, as well as ethical concerns.

“In addition, CAs' adoption can be impeded by communication and interaction problems, as well as ethical concerns (Heyselaar & Bosse, 2019).”

• cites Heyselaar & Bosse, 2019Not checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

The paper discusses the moderating role of big technologies and novel technological advancements like OpenAI GPT or BLOOM NLU that mark the next frontier of CA research.

“Therefore, we discuss the moderating role of big technologies, and novel technological advancements like OpenAI GPT or BLOOM NLU that mark the next frontier of CA research.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Publication venues were cleaned by inspecting and fixing different venues with the same title and spelling variations.

“2) Publication venues were cleaned by inspecting and fixing different venues with the same title and spelling variations (e.g., Computers & Education and Computers and Education).”

Classified as probably not a substantial claim. May combine two claims.

Only journal articles, conference proceedings, or book chapters of conference proceedings published in English prior to 2021 were included; reviews, editorials, and other non-original articles were excluded.

“Only journal articles, conference proceedings, or book chapters of conference proceedings that were published in English prior to 2021 (to include complete years in the trend analysis) were included.Reviews, editorials, and other non-original articles were excluded.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Keywords were compiled and cleaned with the aid of Google Openrefine.

“3) Keywords were compiled and cleaned with the aid of Google Openrefine.”

May combine two claims.

The Scopus database was chosen because it covers most articles included in Web of Science and includes a larger selection of technical and social science articles.

“We chose the Scopus database because it covers most articles included in Web of Science and includes a larger selection of technical and social science articles (Norris & Oppenheim, 2007).”

• cites Norris & Oppenheim, 2007 sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

The final search formula was 'chatbot*' or 'virtual assistant*' or ('conversation* agent*') or 'smart assistant*' or 'smart bot*' or 'natural language interface'.

“The final search formula was “chatbot*” or “virtual assistant*” or (“conversation* agent*”) or “smart assistant*” or “smart bot*” or “natural language interface”.”

Probably not a substantial claim: may not pass S3 (load-bearing).

To avoid assigning higher weights to manuscripts with a higher number of authors, a fractional counting method was chosen to build a weighted co-authorship network.

“To avoid assigning higher weights to manuscripts with a higher number of authors, we chose a fractional counting method to build a weighted co-authorship network (i.e., edge weights are inversely proportional to the number of authors) based on the recom mended methods by Perianes-Rodriguez et al. (2016).”

• cites Perianes-Rodriguez et al. (2016) sourceNot checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

A keyword network was constructed using the full counting method, where keywords in the same manuscript were considered connected.

“A keyword network was constructed using the full counting method, where keywords in the same manuscript were considered connected.”

Classified as probably not a substantial claim.

The ten most cited articles were reviewed to arrive at an understanding of relevant topics driving CA research.

“In the following, we reviewed the ten most cited articles to arrive at an understanding of relevant topics driving CA research (see Table 1).”

• AssumptionThe authors state this as an assumption: The ten most cited articles are substantially relevant for establishing and driving a common understanding of CA as a concept and making CAs prominent in the fields of Human-Computer Interaction and Information Systems.

Classified as probably not a substantial claim. May combine two claims.

A total of 5,107 articles were judged as matching the inclusion criteria, and one article was a duplicate and removed.

“A total of 5,107 articles were judged as matching the inclusion criteria, and one article was a duplicate and removed.”

Probably not a substantial claim: may not pass S3 (load-bearing).

A total of 6,397 articles were retrieved.

“A total of 6,397 articles were retrieved.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The search protocol followed the PRISMA-S method.

“The search protocol followed the preferred reporting items for systematic reviews and meta-analyses literature search extension (PRISMA-S) method (Rethlefsen et al., 2021).”

• cites Rethlefsen et al., 2021 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The cleaned dataset was analyzed using the R statistical language Bibliometrix package.

“The cleaned dataset was analyzed using the R statistical language Bibliometrix package (Aria & Cuccurullo, 2017).”

• cites Aria & Cuccurullo, 2017 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Author names were manually checked and fixed with the aid of Scopus author IDs.

“1) the author names were manually checked and fixed with the aid of Scopus author IDs.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The country collaboration network was constructed in the same way (fractional counting) based on the authors' country affiliation.

“The country collaboration network was constructed in the same way (fractional counting) based on the authors' country affiliation.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Two reviewers reviewed 681 articles and agreed on 656, with an interrater agreement Cohen's Kappa of 0.963 (adjusted = 0.892).

“Two reviewers (first author and third author) reviewed 681 articles and agreed on 656, with an interrater agreement Cohen's Kappa of 0.963 (adjusted = 0.892).”

May combine two claims.

The dispersion of publication dates points toward the importance of the first technology established in this field and the ongoing interest in research on CAs fueled by ongoing advances in related technology.

“This dispersion points toward the importance of the first technology established in this field and the ongoing interest in research on CAs fueled by ongoing advances in related technology.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The diversity of definitions and connotations of CAs illustrates the fragmentation of CA research and lack of agreement on the terminology.

“This diversity also illustrates the fragmentation of CA research and lack of agreement on the terminology.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The resultant network was plotted using Gephi with the Fruchterman Reingold layout algorithm.

“The resultant network was plotted using Gephi (Bastian et al., 2009) with the Fruchterman Reingold layout algorithm (Fruchterman & Reingold, 1991), and each community was assigned a unique color (Fruchterman & Reingold, 1991).”

• cites Bastian et al., 2009Not checked yet

  Not checked in the free run.

• cites Fruchterman & Reingold, 1991Not checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

Subgroups of authors who frequently collaborated were mapped using Louvain modularity.

“Subgroups (i.e., communities) of authors who frequently collaborated were mapped using Louvain modularity (De Meo et al., 2011) to highlight groups who frequently collaborated and their contribution and influence on the field.”

• cites De Meo et al., 2011Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

For readability, the network was limited to the top 100 authors, based on a weighted degree threshold of 10.

“For readability, the network was limited to the top 100 authors, based on a weighted degree threshold of 10.”

May combine two claims.

The Scopus database has rigorously maintained metadata and a curated list of journals and conferences that are subjected to a strict inclusion process.

“The Scopus database has rigorously maintained metadata and a curated list of journals and conferences that are subjected to a strict inclusion process (Baas et al., 2020;Singh et al., 2021).”

• cites Baas et al., 2020 sourceNot checked yet

  Not checked in the free run.

• cites Singh et al., 2021 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The first wave of conversational agents was concerned with achieving a computer program that can resemble a conversation with a human under predetermined specific conditions.

“This first wave of CAs was concerned with achieving a computer program that can resemble a conversation with a human under specific conditions that are predetermined.”

Classified as probably not a substantial claim.

The emergence of conversational agents rooted in the first simple rule-based agents like ELIZA marks the first wave of conversational agent research, the zero hour wave.

“In this regard, the emergence of CAs, which is rooted in the very first, simple, and rule-based CAs like ELIZA, would mark the first wave of CA research -the”

May combine two claims.

The most prominent example from the second wave era is A.L.I.C.E. (release 1995) and its special language, artificial intelligence markup language.

“The most prominent example from that era is the A.L.I.C.E.(release 1995) and the special language it was programmed with, the so-called artificial intelligence markup language (Wallace, 2009).”

• cites Wallace, 2009Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

Although first approaches in conversational agent research existed before 2000, none represented true AI capabilities that first started to emerge about 2005 at least in the domain of conversational agent research.

“Regarding”

Probably not a substantial claim: may not pass S3 (load-bearing).

These technologies allow Rasa to overcome limitations of past frameworks that failed to understand natural language to a large extent and heavily relied on click-based solutions.

“These technologies allow Rasa to overcome the limitations of past frameworks that failed to understand natural language to a large extend and heavily relied on click-based solutions.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

This development highlights the importance of natural language understanding technologies for the present and future of conversational agents and conversational agent research and the quickly and vastly growing capabilities of these technologies that enable next generation conversational agents.

“This development also highlights the importance of NLU technologies for the present and future of CAs and CA research as well as the quickly and vastly growing capabilities of these technologies that enable next generation CAs and thus further pushing the frontiers of CA research.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

With the release of IBM Watson in 2006, IBM became the first big tech company to release a sophisticated product in the domain of conversational agents.

“With the release of Watson, IBM became the first big tech company to release a sophisticated product in the domain of CAs.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Voice-based CAs find little appearance in the keyword cluster, which may be the result of a certain choice of dominant keywords or voice-based CAs not having been researched as extensively as text-based ones.

“Voice-based CAs find little appearance in our keyword cluster, which may be the result of a certain choice of dominant keywords or voice-based CAs not having been researched as extensively as text-based ones.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Nowadays, many people cannot imagine a life without conversational agents like Amazon's Alexa or Apple's Siri.

“Nowadays, many people cannot imagine a life without CAs like Amazon's Alexa or Apple's Siri (Kendall et al., 2020).”

• cites Kendall et al., 2020 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S4 (based).

Considering keyword development over time, topics around true or general AI will significantly gain in importance.

“Considering the development of keywords over time, we predict that topics around”

May combine two claims.

The third wave kick-off is reflected in both the development of conversational agents and related technology and the emergence and distribution of keywords.

“This kick-off is reflected in both the development of CAs and related technology (e.g., frameworks, implementations) and the emergence and distribution of keywords.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

The second wave of conversational agent research presented more advanced agents that for the first time made use of natural language processing and statistical methods.

“The second wave of CA research presented us with more advanced CAs that, for the first time, made use of NLP and statistical methods.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

AI-driven intelligent embodied conversational agents may become more popular and, with increasing advancements, blur the lines between conversational agents and robot artifacts.

“For instance, AIdriven intelligent embodied CAs may become more popular and, with increasing advancements, blur the lines between CAs and robot artifacts (Huang & Rust, 2021).”

• cites Huang & Rust, 2021 sourceNot checked yet

  Not checked in the free run.

May combine two claims.

Overall, machine learning and AI seem to be two of the core topics of the current wave of conversational agent research and current generation conversational agents as well as novel conversational agent artifacts.

“Overall, machine learning and AI seem to be two of the core topics of current wave of CA research and the current as generation CAs as well as novel CA artifacts (Suta et al., 2020).”

• cites Suta et al., 2020Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

Topics like embodiment and virtual agents, including avatars and other anthropomorphic approaches, may increase in significance as the ability to model human-like digital artifacts develops.

“In addition, topics like embodiment and virtual agents, including avatars and other anthropomorphic approaches, may increase in significance as our ability to model human-like digital artifacts develops.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

By examining past research, CA research has undergone certain incremental steps that most likely have been enabled by technological advancements.

“By examining past research, we can observe that CA research has undergone certain incremental steps that most likely have been enabled by technological advancements.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The development of generative AI and large language models brings in beta versions of the future of conversational agents.

“Especially the development of generative AI and according large language models bring in beta versions of the future of CAs.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Google LaMDA provides new capabilities for natural language interactions through training a transformer model on conversations.

“Current examples include for instance the Google conversational technology”

• cites Thoppilan et al., 2022Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

First conversational agents appeared in the second wave that could somewhat understand and resemble human emotions via scripted dialogue.

“Furthermore, first CAs appeared that could somewhat understand and resemble human emotions via scripted dialogue.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

The emergence of multimodality in the third wave highlights that voice-based and multimodal conversational agents would come after and are omnipresent today.

“The emergence of this keyword at that time highlights that what should come after, which is what we see to be omnipresent today voice-based and multimodal CAs.”

May combine two claims.

Because of technological limitations, the second wave of conversational agents never managed to overcome the uncanny valley, marking a significant gap in research and practice that in some areas persists until today.

“Although best efforts to humanize CAs were made (Porra et al., 2020), because of technological limitations, this wave of CAs never managed to overcome the uncanny valley of CAs, thus marking a significant gap in research and practice that in some areas persists until today.”

• cites Porra et al., 2020 sourceNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Rasa can be built to include spaCy and BERT for its natural language pipeline.

“In this regard, Rasa can be build to include spaCy (Honnibal et al., 2020) and BERT (Devlin et al., 2018) for its' natural language pipeline.”

• cites Honnibal et al., 2020Not checked yet

  Not checked in the free run.

• cites Devlin et al., 2018Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Widespread adoption of these developments would potentially enable conversational agents to become even more humanlike, to the point where humans may mistake a conversational agent for a real human.

“A widespread adoption of these developments would potentially enable CAs to become even more humanlike, to the point where humans may mistake a CA for a real human as in the above introduced example of Google LaMDA (Thoppilan et al., 2022).”

• cites Thoppilan et al., 2022Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

The first wave had strict technical limitations due to the computational capabilities of that era and therefore could not produce more advanced artifacts.

“The first wave had strict technical limitations due to the computational capabilities of that era and therefore couldn't produce more advanced artifacts.”

May combine two claims.

The keyword deep learning first emerged in conversational agent research after 2015 and emphasizes the potential AI-route conversational agent research may take in following years.

“Particularly, the keyword”

Probably not a substantial claim: may not pass S3 (load-bearing).

The keywords AI, NLP, and machine learning showcase what future road conversational agent research may take.

“Moreover, from a more technical point of view, the keywords”

May combine two claims.

Although the second wave brought first attempts at AI in conversational agent research, AI itself was still in a very early stage due to technological limitations and the virtual non-existence of voice-based conversational agents.

“However, although this wave brought first attempts at AI in CA research, AI itself was still in a very early stage due to technological limitations and the virtual non-existence of voice-based CAs.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

Topics about true AI include full automation and autonomous conversational agents, which may be enabled by future advancements toward self-learning and real-time adaptable AI applications.

“These topics about true AI include full automation and autonomous CAs, which may be enabled by future advancements toward self-learning and real-time adaptable AI applications.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The release of the beta version of OpenAI's ChatGPT based on further developments of GPT at the end of November 2022 is called by experts the most disruptive AI technology in 2022 and the future.

“Most remarkable is the release of the beta version of OpenAI's ChatGPT based on further developments of GPT (Stokel-Walker, 2022;Brown et al., 2020b) end of November 2022, called by experts as the most disruptive AI technology in 2022 and the future (Haque et al., 2022).”

• cites Stokel-Walker, 2022Not checked yet

  Not checked in the free run.

• cites Brown et al., 2020bNot checked yet

  Not checked in the free run.

• cites Haque et al., 2022Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

In the 2010s, many advanced conversational agents enabled by significant advancements in AI and natural language processing became mainstream and omnipresent in everyday lives.

“In the 2010s, many advanced CAs enabled by significant advancements in technology, particularly in AI and NLP, became mainstream and even omnipresent in our everyday lives.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

The actual understanding of natural language and the intentions behind it may become more important and powerful, based on advanced deep learning and natural language understanding methods.

“Moreover, the actual understanding of natural language and the intentions behind it (i.e., what humans mean vs. sarcasm) may become more important and powerful, based on advanced deep learning and NLU methods (Suta et al., 2020).”

• cites Suta et al., 2020Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing), S4 (based).

The keywords chatbot and conversational agent highlight the real hype conversational agent research received from 2015 onwards.

“The keywords”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

This led to a vast increase in popularity and subsequently further development, research, and financial investment into conversational agents, as researchers and big tech players like Google or Amazon realized the potential.

“This in return led to a vast increase in popularity and, subsequently, further development, research, and financial investment into these agents, as not only researchers but also big tech players like Google or Amazon realized the potential.”

May combine two claims.

The second wave led to dedicated languages and somewhat more advanced conversational agents entering the field.

“This led to dedicated languages, and somewhat more advanced CAs entering the playing field.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

The emergence of conversational agents focused on crude artifacts with predetermined and programmed structure.

“Circling back to the very beginning of CAs, we can observe that the emergence of CAs focused on rather crude artifacts with predetermined and programmed structure.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Conversational agents enjoyed a real hype from academia to corporate that swept over into the real world.

“CAs enjoyed a real hype from academia to the corporate that swept over into the real world.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

The release of Watson started to draw significant attention from other big-tech companies in the following years.

“That development started to draw significant attention from other big-tech companies in the following years.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

First, AI lite conversational agents emerged that made use of more than simple rule-based, statistical, or light AI methods.

“First, AI”

May combine two claims.

The first embodied conversational agents with anthropomorphic features emerged and became somewhat popular in the second wave.

“In addition to these developments, the first embodied CAs with anthropomorphic features emerged and became somewhat popular.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Interfaces combining sophisticated conversational agent interfaces such as ChatGPT with avatar generators could enable fully customizable conversational agents with unmatched social presence and anthropomorphism.

“Interfaces building upon machine learning pipelines that combine sophisticated CA interfaces, such as ChatGPT (Stokel-Walker, 2022), with avatar generators could enable in the future fully customizable CAs with a previously unmatched degree of social presence and anthropomorphism.”

• cites Stokel-Walker, 2022Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

While a system's behavior based on machine learning was largely shaped by its input data, large language models introduce a novel facet of opacity to the user.

“While a system's behavior based on machine learning was largely shaped by its input data, large language model introduce a novel facet of opacity to the user.”

Classified as probably not a substantial claim.

The authors expect to see a shift of research emphasis from trust in the system towards trust in the provider.

“As a result, we expect to see a shift of research emphasis from trust in the system towards trust in the provider.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

Considering how past technological advancements have been used by corporate players and monetized, the technology may go proprietary and paid.

“Considering how past technological advancements have been used by corporate players and monetized, this may raise the concern of the technology going proprietary and paid.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

Strategic decisions by corporate players might have larger implications on the conversational agent landscape, including organizations deciding whether to make or buy generative AI-based conversational agents.

“Such strategic decisions might have larger implications on the CA landscape, including organizations deciding whether to”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

OpenAI is not only interested in developing frontier level technology for conversational agent research but also in monetizing it.

“However, looking closer at OpenAI it is evident that this corporate player is not only interested in developing frontier level technology for CA research but also in monetizing it.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

AI and relevant topics related to it, such as trust in AI and explainable AI, are very likely to continue throughout the next wave.

“Here, we find AI and relevant topics related to it (e.g., trust in AI, explainable AI) to be very likely to continue the throughout the next wave.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

Questions of how a conversational agent's underlying model behaves and affects users become even more important.

“Questions of how a CA's underlying model behaves and affects users become even more important.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The research was conducted according to established research guidelines and used established statistical methods followed as rigorously as possible.

“Although we conducted our research according to established research guidelines and used established statistical methods that we followed as rigorously as possible, our research may not be without limitations.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The authors expect voice-based conversational agent research to become more popular and potentially become the modality of choice in the future.

“As explained in our five-wave view on CA research, this observation reflects a natural development in CA research, and we expect voice-based CA research to become more popular and potentially become the modality of choice in the future (e.g., Rzepka et al. (2021)).”

• cites Rzepka et al. (2021)Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

ChatGPT is trained on a large language model with a cut-off point at the end of 2021.

“For instance, the ChatGPT (Stokel-Walker, 2022) CA is trained on a large language model with a cut-off point end of the year 2021.”

• cites Stokel-Walker, 2022Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

Technological developments of conversational agent research and advancements in related fields like AI have raised the bar for technological requirements and are likely to continue this trend.

“First, the technological developments of CA research and also advancements in related fields like AI have raised the bar for technological requirements and are likely to continue this trend.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The analyses on commercially-funded publications could not consider all potential sources and should be interpreted with caution.

“Although our analyses on commercially-funded publications could not consider all potential sources and thus should be interpreted with caution, they provide a first glimpse into emerging sectors.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Data allows AI to deliver AI-based core functionality including improved natural language processing and understanding capabilities but also adaptability.

“Data not only allows AI to deliver AI-based core functionality including improved natural language processing and understanding capabilities but also adaptabiliy.”

Classified as probably not a substantial claim.

Future research may likely see a strong focus on data-driven and AI-driven approaches, as highlighted by keyword analysis and the five-wave view.

“Here, we have pointed out that future research may likely see a strong focus on data-driven and AI-driven approaches, as highlighted by our keyword analysis and five-wave view on CA research.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Users' expectations regarding the capabilities of conversational agents will increase and the need for explainable AI and transparency potentially decrease, especially from a user's perspective.

“At the same time, users' expectations regarding the capabilities of CAs' will increase and the need for explainable AI and transparency potentially decrease, especially from a user's perspective.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The authors may have missed some niche domain or research stream despite choosing appropriate keywords and outlets and not using publication year restrictions.

“First, although we did our best to include as many relevant articles as possible by choosing the appropriate keywords and outlets and not using limitations like publication year restrictions, we may have missed some niche domain or research stream.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

With regard to individualization and personalization of conversational agents, AI is only one relevant aspect; another relevant aspect is the design of such AI-driven artifacts.

“With regard to individualization and personalization of CAs, AI is only one relevant aspect, another relevant aspect is the design of such AI-driven artifacts (Behera et al., 2021).”

• cites Behera et al., 2021Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

Reviewing the articles with more or other reviewers may have led to different results and a somewhat divergent body of literature.

“Reviewing the articles with more or other reviewers may have led to different results and, thus, a somewhat divergent body of literature.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The authors may not have included literature that uses more specific keywords exclusively, such as smart personal assistant or agent.

“Moreover, we may not have included literature that uses more specific keywords exclusively.An example of this would be research that uses”

May combine two claims.

Underlying technical papers on large language models such as GPT were previously published, but technologies such as ChatGPT now remain proprietary.

“This is right now also visible through the development that underlying technical papers on large language models such as GPT were previously published (Brown et al., 2020b) but technologies such as ChatGPT now remain proprietary.”

• cites Brown et al., 2020bNot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

Moving from niches of computer science to standard technology, users' expectations regarding conversational agents' capabilities will continuously increase.

“Moving from the niches of computer science to standard technology, users' expectations regarding CAs' capabilities will continuously increase.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

Advancements in technological models and understanding thereof have made contemporary conversational agents sophisticated and commercially available to the large public.

“Advancements in technological models and understanding thereof have made contemporary CAs sophisticated and commercially available to the large public.”

Probably not a substantial claim: may not pass S3 (load-bearing).

This has different implications for each type of conversational agent as well as the types of interaction users can have with conversational agents.

“This, however, has different implications for each type of CA as well as the types of interaction users can have with CAs (e.g., text-based, voice-based, embodied) (Nguyen et al., 2021).”

• cites Nguyen et al., 2021Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The increasing appearance of keywords like affective computing and emotion reflects efforts to make conversational agents more human-like by touching on interpersonal human-human interactions.

“This is also reflected in the increasing appearance of keywords like”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

Interacting with conversational agents will become even more easily accessible, effortless, and integrated in humans' daily personal and work life.

“In turn, interacting with CAs will become even more easily accessible, effortless, and integrated in humans' daily personal and work life.”

May combine two claims.

These findings either indicate the relevance of certain conversational agents for particular domains or point toward the fragmentation of conversational agents across research outlets.

“These findings either indicate the relevance of certain CAs for particular domains or point toward the fragmentation of CAs across research outlets.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

This could become a hindrance in further advancing the frontiers of conversational agent research as many researchers or practitioners may be disinclined to comply with these terms.

“This could become a hindrance in further advancing the frontiers of CA research as many researchers or practitioners may be disinclined to comply with these terms.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Future developments might include interactions based not only on historical data but also real-time data.

“Future developments might include in this context that interactions are not only based on historical data but also real-time data.”

Probably not a substantial claim: may not pass S2 (checkable).

Considering the interest of big corporate techs in AI and related topics that oftentimes act as affiliates of conversational agent research, this development will be at the core of the fifth wave of conversational agent research toward true AI-driven conversational agents.

“Considering the interest of big corporate techs in AI and related topics that oftentimes act as affiliates of CA research, we predict this development to be at the core of the fifth wave of CA research toward true AI-driven CAs.”

• AssumptionThe authors state this as an assumption: The development of AI-related keywords represents a novel hype of AI-related technology that will grow even further in the future.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

Underlying functions and interrelatedness within a larger commercial ecosystem are becoming increasingly opaque.

“In particular, the underlying functions and interrelatedness within a larger commercial ecosystem are becoming increasingly opaque.”

Probably not a substantial claim: may not pass S3 (load-bearing).

With increasing interest of Chinese players such as Baidu, there may be a shift toward Chinese affiliations during the next years.

“Furthermore, with the increasing interest of Chinese players (e.g., Baidu), we may see a shift toward Chinese affiliations during the next years.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Only two of the authors reviewed the literature in detail and made decisions to include or exclude the literature.

“Second, only two of the authors reviewed the literature in detail and made decisions to include or exclude the literature.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

Recent developments around generative AI and large language models illustrate that conversational agents and their technical sophistication are a moving target, just like AI.

“Recent developments around generative AI and LLM illustrate that, just like AI, CAs and their technical sophistication are a moving target.”

Probably not a substantial claim: may not pass S4 (based).

The authors expect questions of how conversational agents will behave and their implications for how humans interact, work, and rely on external advice to be more important than ever.

“Last, we expect the questions of how CAs (will) behave and their implications for how humans interact, work, and rely on external advice, to be more important than ever.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Public omnipresence has introduced tech corporations into the academic research field, with both becoming increasingly intertwined.

“Public omnipresence, in turn, has introduced tech corporations into the academic research field, with both of them becoming increasingly intertwined.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

Emergent fields of application for CAs, as well as transforming existing fields and pushing boundaries, may become even more significant with increased societal impact.

“Emergent fields of application for CAs as well as transforming existing fields and pushing the boundaries may become even more significant with increased societal impact”

• cites Wahde & Virgolin, 2022Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

CAs already have a significant impact on society, which is expected to extend.

“CAs already have a significant impact on society which is expected to extend.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

Restriction of access to CA software may have serious adverse effects and hinder development and contributions by independent researchers and developers.

“restriction of access may have serious adverse effects and hinder development and contributions by independant researchers and developers.”

• AssumptionRests on an assumption: Free and open-source CA software fosters CA research and enables independent researchers to contribute.