Which of the following medium of communication has the highest information richness?

Relationship Maintenance Strategies and Communication Media

Media richness theory or information richness theory, developed by Daft and Lengel (1986), was an early framework for maximizing effectiveness of communication within the field of organizational management. The theory was developed prior to the existence of many of the electronic applications available to today's NPRRs. However, basic concepts that began in media richness theory find a place in current research. From the perspective of media/information richness theory, richness of media communication is determined by its ability to efficiently improve understanding or learning. Goals of communication are to improve certainty and decrease ambiguity in order to build understanding and overcome differences in perspectives (Daft & Lengel, 1986). Media used for communication varies in ability to convey complex information using multiple cues. Face-to-face communication is the best vehicle for conveyance of complex information including facial expression, gestures, and tone of voice, followed by telephone conversations which also allow for nuances in tone of voice, pauses, subtle emphases, etc. Text-based interactions allow for less richness in cues (Daft & Lengel, 1984). Today, electronic devices offer enhanced opportunities to enrich communication when partners in NPRRs are unable to be in the same location for extended time periods.

Relationship satisfaction for couples in NPRRs relies a great deal on technology-mediated communication and is dependent on which communication channels are most frequently used. Dainton and Aylor (2002) found that the use of telephone and Internet were positively correlated with relationship satisfaction for NPRRs. However, this study did not include more recent communication technology channels such as Facebook, Instagram, Twitter, or Skype.

Johnson, Haigh, Becker, Craig, and Wigley (2008) examined how college students used email to maintain interpersonal relationships with friends, family, and romantic partners; types of messages sent; and whether distance was a moderator. They found that distance tended not to moderate relationship maintenance strategies. Categories of email exchanges were similar across different types of relationships, with one exception: Romantic partners engaged in more exchanges of open, direct discussion characterized by give-and-take interactions.

Expanding on research by Johnson et al. (2008) that identified email as the preferred communication medium, Kirk (2013) examined ways in which newer technologies affect NPRRs, specifically relationship satisfaction and maintenance strategies. Skype was found to be the most preferred computer-mediated communication channel, compared with Facebook, email, or Twitter. Time spent on Skype video chat was positively correlated with relationship satisfaction. Skype is a multifaceted communication tool that provides a means for relationship maintenance strategies. Relationship maintenance strategies, such as open, direct discussion and positive and upbeat statements are frequently used by NPRR couples and the use of maintenance strategies in computer-mediated communication is positively related to communication satisfaction (Rabby, 2007; Wright, 2004).

For relationships of all kinds, choice of interpersonal media is influenced by circumstances specific to the relationship, such as distance, as well as customary behaviors within the relationship and the amount of desired control over the communication (Rabby & Walther, 2003 as cited in Johnson et al., 2008). Media choice is a particularly salient issue for couples in NPRRs because it requires that the communicator strategize about which type of media to employ to best achieve the desired interpersonal outcome. Jiang and Hancock (2013) posit that the decision about which form of media communication to use is based on three media variables: synchronicity (Walther, 2007), mobility (Dimmick, Feaster, & Hoplamazian, 2011), and cue multiplicity (Daft & Lengel, 1986).

Synchronicity refers to real-time communication. Synchronous media includes messages exchanged instantaneously in real time, allowing for simultaneous interactions (e.g., face-to-face, video chat, and phone call). Text messaging, instant messaging (IM), and email are considered asychrononous media, given the potential for even brief response latency (Walther, 2007).

Mobility relates to media portability. For example, speaking or texting via mobile phone offers the highest media mobility (Dimmick et al., 2011). Computer-mediated communication such as email, video chat, and IM provide some mobility, especially with the growing number of tablet, laptop, and smartphone users but their use is less facile than that of mobile phones.

Cue multiplicity refers to the extent to which a media device can simultaneously convey multiple communication interaction cues, such as voice inflection, facial expression, body movement, and verbal expression (Daft & Lengel, 1986; Dennis, Fuller, & Valacich, 2008). The simultaneous, electronic exchange of verbal, audio, and visual cues is most effectively accomplished via video chat. Written or typed communication represents decreased cue multiplicity. The need for cue multiplicity, synchronicity, and mobility varies depending on the complexity, content, and urgency of the communication need, with more complex interpersonal interactions requiring high cue multiplicity, high synchronicity, and less mobility (Jiang & Hancock, 2013).

In the pursuit of a desired outcome, such as intimacy enhancement, partners will adapt communications to fit the constraints of nonproximal relationships. The mode of interpersonal media enhances or restricts communication in specific ways. Communicators cognitively assess how many communication cues are available, whether they can interact in real time, and whether cues are accessible if the communicator is moving from one location to another during the communication (Jiang & Hancock, 2013).

Intimacy and interpersonal dynamics operate in distinct patterns and vary greatly across communication media. For example, NPRR couples use different maintenance strategies depending upon the type of interpersonal media, such as phone calls, video chat (Skype), IM, texting, electronic mail (email), and social media forums (Facebook; Dainton & Aylor, 2002; Jiang & Hancock, 2013). In examining categories of relationship maintenance behaviors (described by Canary & Stafford, 1994), Dainton and Aylor (2002) found that positivity (upbeat and pleasant interactions), social networking (relying on support of friends and family), and shared tasks (routine relationship chores) seem to be emphasized more in text-based communications, whereas assurance of affection and openness (direct, give-and-take discussions) are emphasized on the telephone. For instance, those insecure in their relationship may prefer verbal communication (to see or talk to partner) for reassurance due to a desire for immediate responses and voice cues that may be lacking or limited within written communication channels (Dainton & Aylor, 2002).

Jiang and Hancock (2013) found that when couples use text-based, asynchronous, and mobile media for communication, information exchange is narrowed by fewer cues and less immediacy. Couples compensated for fewer cues and lack of immediacy in their communications by increasing the number of self-disclosures and idealizing or overemphasizing their partner's disclosures which served to build intimacy. For example, a brief romantic message (e.g., “thinking of you”) may be saved and revisited, serving as a symbol of connectedness, which further enhances positive emotions. This finding was supported by Stafford's (2010) report that when comparing couples in PRRs with others in NPRRs, couples in NPRRs reported more intimacy in their communications and activities which may be accentuated by higher levels of conflict avoidance and more limited discussion about making decisions that move toward marriage.

Jiang and Hancock (2013) results also revealed that partners in NPRRs engaged in fewer interactions per day than proximal participants but reported using more media such as phone calls, video chats, texting interactions, and IM interactions. Both proximal and nonproximal couples reported minimal email communication compared with earlier studies (Dainton & Aylor, 2002; Johnson et al., 2008).

Information richness.

The opportunity to compile different types of content and content of different quality. Similar to the ‘media richness’ feature, the information richness enables companies to present their content in different service levels and for different target groups, from which additional business opportunities may be generated. Yet, as there is practically no limit with respect to the amount of information, which can be presented on a corporate Web site, companies can try to compile all of the information which they believe to be of interest for their customers on their Web site. So far, this is not possible for any other communications medium. For example, portal sites follow this approach, trying to become their visitors’ single point of access for the Worldwide Web and dominate their visitors’ whole online experience. Less ambitious companies in manufacturing or high technology might post all of their product catalogues, technical specifications, data sheets and support information on their Web sites.

V.E. From Information Poverty to Information Richness

Information “poverty” characterized the early stages of MIS. Information was difficult to obtain, untimely, and expensive. Sales data were obtained, laboriously compiled, and made available to managers on a monthly basis. Financial performance data typically were summarized quarterly. Indeed, it was the desire to have information available in a more timely fashion and at lesser cost that led to the need for, and creation of, MIS.

The current era is characterized by information richness—the availability of information in real time and at low cost. Indeed, some might characterize the situation as one of “information overload” if not for the development of inquiry-based systems which permit more information to be available on an “as needed” basis rather than solely in the form of regularly issued reports.

These changes have revolutionized management, making it possible to track transactions instantaneously (as in financial markets) and to acquire information at a micro level of detail, such as the “stream” of supermarket purchases of specific brands, sizes, and varieties of products by individual consumers. The systems that have been enabled by the availability of such data range from automated programmed trading systems for the purchase and sale of financial instruments to “frequent purchaser” programs for supermarkets that offer discounts, special incentives, and other benefits to their best customers, to customer relationship management which is facilitated by accumulating a wealth of information about a customer, his purchasing habits, and his purchasing search behavior.

2.01.4.4 Data and GIS Methods for Enhanced Land Use/Land Cover Mapping

Although the usefulness of GIS concepts for LULC mapping has been known for decades, they are—to date—comparatively seldom exploited to the fullest in the LULC mapping community. However, from a GIS perspective, methods applied after the classification process can even become the dominant part of the LULC map production. Nowadays, popular remote sensing software such as ENVI also contain basic GIS functionalities. Nevertheless, matured GI-Systems contain much more powerful overlay or data integration tools for the handling of categorical information (cf. Heywood et al. (2011); Longley et al. (2015)).

Furthermore, manifold sophisticated geodata are available for the different scale levels, which already contain high-quality information for many LULC categories (see previous section, section “Vegetation Data in Official Information Systems”). Good examples for the information richness, such datasets are vector-based DLMs, which are usually intended for the scale range of 1:10,000 to 1:25,000. In general, such datasets can be used for regional to nation-wide LULC mapping. DLMs not only provide basic topographic information, they also contain a lot of land use information. Nowadays, DLMs are available as open data for many countries. Example datasets are the Topologically Integrated Geographic Encoding and Referencing (TIGER) data for the USA (www.census.gov), the Authorative Topographic-Cartographic Information System (ATKIS) for Germany (AdV, 2006) or the TOP10NL for the Netherlands (Kadaster, 2016). In addition, for very high spatial scales (i.e., ca. 1:1000 or larger) cadastral data like the Authoritative Real Estate Cadastre Information System (ALKIS) for Germany (AdV, 2016) is available.

The ATKIS, for example, describes the landscape in the form of point line or polygon features. Due to its origin from the official state survey and mapping agency, it is characterized by a high geometric accuracy (AdV, 2006). The ATKIS includes diverse information, for instance on the transportation network, but it also differentiates built-up areas into different types of residential or industrial areas, forests into deciduous and coniferous tree populations as well as agricultural areas into arable land, grassland or orchards. Fig. 8 shows a selection of the ATKIS (Basis DLM) information content occurring in this area.

The categorical information (cf. Table 1) in such datasets, which indicates the actual land use, is especially valuable. In this regard, the ATKIS conveys detailed information, if a certain area is (e.g.) a golf course or an airfield, while the basic land cover may be grassland in both cases. Other spatial information like the boundaries of protection areas may also help to narrow down the actual land use or the land use intensity (Bareth and Waldhoff, 2012). As stated earlier, much of this information cannot be obtained from remote sensing at all, or not in the provided geometric and thematic quality. The combination of data and methods available in GIS environments can therefore substantially enrich the information content, especially concerning land use, and even increase the geometric quality of LULC maps.

4.2 Media Richness Theory

One of the most widely applied theories of media selection is Media Richness theory by Daft and Lengel [10, 11]. Media Richness, which builds on the theory of Social Presence, argues that communication media differ in their ability to facilitate understanding. Daft and Lengel have defined information richness as the capacity of information “to change understanding within a time interval” [9]. Thus, in Daft and Lengel's terms, what differentiates richer media from leaner media is the amount of information a medium could convey to change the receiver's understanding within a time interval. This capacity depends on several factors, such as the ability of the medium to transmit multiple cues, immediacy of feedback, and language variety. The perceived sense of social presence of a medium is proportional to the medium richness. As a result, rich media with a wide communication capacity also have a high level of social presence. F2F interaction is the richest media, due to its capability of expressing message context in natural language and conveying at the same time multiple cues via body language and tone of voice, and it is supposed to change understanding of participants in communication in a shorter time interval. The second richest medium is videoconferencing, because, although it still grants the use of natural language, and the access to some visual and verbal cues, it conveys a lower sense of social presence to conversation participants. E-mail, chat/IM, and letters are instead the leanest media because, when adopted, communication exchanged by participants is conveyed on a single channel, that is, text, be it written or typed.

Like Social Presence Theory, also Media Richness theory presumes that the outcome of an interaction is determined by the communication capacity of the selected medium. While Social Presence theory relates performance primarily to the type of interaction required (relational vs. activity focused), Media Richness Theory asserts, instead, that performance depends on the appropriateness of the match between media richness characteristics and information requirements of the task (clarification vs. additional information). Indeed, Media Richness Theory postulates the existence of two complementary forces that act on participants when they process the information exchanged when communicating (see Fig. 7). One force is uncertainty, which is defined as the “difference between the amount of information required to perform a task and the amount of information already possessed” [11]. This definition builds on earlier research work about information theory (i.e., as information increases, uncertainty decreases [62]). Uncertainty is reduced obtaining additional data and seeking answers to explicit questions. The other force is equivocality, which is the existence of multiple and conflicting interpretations about a situation [11]. As uncertainty is more related to the amount of information available, equivocality is more related on the quality of information available: Equivocality means ambiguity and reflects confusion and lack of common understanding, whereas uncertainty means the absence of sufficient data necessary and reflects the inability to process information properly.

Equivocality is reduced by seeking for clarification, reaching agreement, and deciding what questions to ask. The postulation of the existence of these two complementary forces has also implications on the selection of the most effective medium to use. Media Richness theory posits that rich media are better suited in equivocal communication situations (where there are multiple, even conflicting, interpretations for available information), whereas lean media are best suited in uncertain communication situations (where there is a lack of information). Equivocality is often symptomatic of disagreements and, thus, providing sufficient clarifications can reduce it. Rich media interaction (e.g., F2F), is preferred in situations of equivocality, as it allows for rapid feedback and multiple cues, thus facilitating the convergence to a shared interpretation. On the other hand, when messages are not equivocal, lean media are preferred. Thus, uncertainty can be reduced by obtaining sufficient additional information using media like e-mail or written reports. Therefore, in short, Media Richness proposes that task performance will be improved when tasks needs are matched to the medium ability of conveying information.

Finally, we notice that Daft and Lengel have treated equivocality and uncertainty as independent constructs. However, it must be pointed out that a new amount of data can also generate ambiguity, and that equivocal scenarios may need more data to converge as well.

2.3.12 AffinityRank algorithm

The AffinityRank algorithm exploits diversity and information richness of a webpage in ranking webpages as looking only for relevance might actually reduce the coverage of the returned results [130]. The algorithm works just like the PageRank algorithm but instead of explicit hyperlinks it uses the similarity among webpages. A link between two webpages is created if their similarity score is greater than a certain predefined threshold. The information richness is based on the intuition that a webpage is more informative if it has many linked pages and informative neighbors. The diversity is obtained using the cosine similarity among the documents. The algorithm produces relevant, rich and diverse search results. However, it uses only the statistical properties of a webpage for document generality [131].

4.3.1 Diversification analysis

To answer question 1, we compare our L-SRD with other diversification models in terms of diversification metrics. Table 1 shows the result of the evaluation in terms of α-nDCG, ERR-IA, and NRBP. The best result per line is highlighted in bold. The classical MMR method is used as a representative of implicit diversification model (Carbonell & Goldstein, 1998). As for explicit model, we consider xQuAD and PM2 (Dang & Croft, 2012; Santos et al., 2010a). SVMDIV is selected for the representative of learning methods, HxQuAD is selected for the hierarchical model.

Table 1. Diversification performance using the official evaluation metrics for WT2009, WT2010, WT2011.

The result shows that L-SRD always performs best in terms of all metrics as shown in Table 1. It consistently improves the initial retrieval ranking method with gains up to 23.19%, 31.17%, 15.11% in terms of α-nDCG on WT2009, WT2010, WT2011, respectively. It indicates that our learning approach tackles the diversity measurement problem more effectively with the consideration of integrate different features. The reason is that features such as query-aspects relevance and information richness conform to the property of diversity. Furthermore, comparing with the explicit diversification models in terms of the evaluation of α-nDCG, the improvement of L-SRD over the xQuAD is up to 28.44%, 16.87%, 14.90% on WT2009, WT2010, WT2011 respectively, and the improvement of L-SRD over the PM2 is up to 14.15%, 23.11%, 14.65% on WT2009, WT2010, WT2011, respectively. Previous explicit diversifications use a predefined function to calculate the diversity score, which cannot reach an optimal result from the overall situation. A learnable approach to solve the diversity measurement and parameter tuning problem is significative. In addition, comparing with the hierarchical diversification model in terms of the evaluation of α-nDCG, the improvement of L-SRD over the HxQuAD is up to 3.77%, 9.68%, 5.49% on WT2009, WT2010, WT2011, respectively. HxQuAD only use a predefined function to measure the diversity score, and the parameters may not be optimal because it needs to be tuned manually. Our learning model tackles the parameters tuning problem in an automatic fashion and reaches optimal result. Comparing with SWME in terms of the evaluation of α-nDCG, the improvement of L-SRD over the HxQuAD is up to 3.63%, 5.35%, 3.86% on WT2009, WT2010, WT2011, respectively. SMWE mines enough subtopics, but it cannot learn enough features to represent the document. Besides the non-learning model, the improvement of L-SRD over the SVMDIV is up to 10.18%, 14.70%, 11.09% on WT2009, WT2010, WT2011, respectively. It shows that considering relevance and different types of features in diversity measurement is helpful in the learning approach. That is the reason why our model wins. Therefore, L-SRD shows better understanding on the diverse ranking and leads to a better result. So we find that utilizing learning mechanism indeed promotes the performance of search result diversification.

We consider not only the advanced diversity metrics, but also traditional diversity metrics, such as Precision-IA and Aspect Recall. The former indicates how many relevant documents for each aspect we have for reranking, the latter indicates how many of the aspects for which we have relevant documents. The result is shown in Fig. 3. MMR still underperforms all of them, as for Precision-IA, xQuAD wins on WT2010 casually, while L-SRD performs more stable, even on WT2010, the gap is small. It proves that L-SRD outperforms others from different perspectives. Our learnable model solves the diverse ranking problem in a global perspective and always reaches prominent results.