Automatic Route-based Ridesharing on the Web

Embed Size (px)

Citation preview

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    1/16

    dinoride 1

    dinoride: Automatic Route-based Ridesharing on the Web

    Benjamin Carl Mearns

    University of Pennsylvania

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    2/16

    dinoride 2

    Abstract

    In the United States, interest in intercity automobile ridesharing abounds, as evidenced by

    the large number of dedicated on and offline resources. Contemporary domestic travelers face

    significant incentives to share rides with others passing along similar routes, amidst mounting

    concern over inflating fuel prices, global warming, oil dependency, traffic congestion, and

    airborne terrorism. This brief paper describes a web-based application, dinoride, which aims to

    fill the need for a web-based centralized, automated, user-friendly rideshare application. The

    rapid growth in availability of structured data -- such as the Craigslist RSS used by Dinoride --

    has made it possible for the urban planning professional to implement user-focused solutions

    leveraging vast sources of information, hitherto unavailable. Dinoride is unique in the tiny but

    fast growing field of web-based ridesharing for its ability to access this type of information, as

    well as for its implementation of a spatial algorithm better suited to the American ridesharing

    landscape: relatively sparse residential settlement, many alternate routes, few available shares,

    and long distances of travel. Dinoride will benefit further from refinement of routing and

    matching rules, expansion of the data model, and extension of the user interface.

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    3/16

    dinoride 3

    dinoride: A Dynamic Web Ridesharing Application

    In the United States, interest in intercity automobile ridesharing abounds, as evidenced by

    the large number of dedicated on and offline resources. Contemporary domestic travelers face

    significant incentives to share rides with others passing along similar routes, especially amidst

    mounting concern about inflating fuel prices, global warming, oil dependency, traffic congestion,

    and airborne terrorism. However, despite both high demand for connecting parallel travelers,

    and booming social networking via the internet, available solutions do not fit the criteria of

    centralized, automated, and user-friendly. This brief paper describes a web-based application,

    dinoride (http://www.dinoride.com), which aims to fill an apparent public need.

    Review of Context

    In a previous essay (Mearns, 2007), I explore the range of contemporary rideshare

    resources from low-tech cork ride boards to high-tech vehicle-integrated dynamic routing

    systems and demonstrate that web based ridesharing systems have been woefully overlooked.

    However, that survey missed a growing list of sites which offer dynamic ridesharing including

    511.org in the San Francisco Bay area (http://www.511.org), Hitch in New Zealand

    (http://www.hitch.net.nz/)and Jack Bell Rideshare in Canada (http://online.ride-share.com/).

    While each of these applications have their own merits, they are neither adequate for typical

    American intercity travel nor long distance commutes, which offer many re-routing opportunities

    incorporating actual share locations.

    http://www.hitch.net.nz/http://www.hitch.net.nz/
  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    4/16

    dinoride 4

    Furthermore, the relatively small number of willing shares for an American intercity trip

    or long-distance commute necessitates a centralized repository, allowing the user to take

    advantage of the widest available rideshare information. No existing application has

    incorporated external information, rather depending upon their separate, limited, user bases.

    While external information has much potential for increasing the utility of ridesharing

    applications, gathering and processing this spatial data requires novel strategies for achieving

    uniformity and determining relevance.

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    5/16

    dinoride 5

    Description of Application

    Dinoride is a system of complimentary scripts and databases consumed through a typical

    dynamic web page. Therefore, among the technologies used are those which have implications

    for the web server, database, and browser domains. PHP (http://www.php.net/documentation/),

    residing within the web server, is a scripting language which allows automated interaction

    between the database and browser domains. One can think of PHP as the script in an interactive

    play in which the web server is the principal actor. When the audience arrives perhaps they

    throw rotten tomatoes at the web server, our principal actor. It is the PHP, in which language the

    web servers script is written, which tells it that it must react to the rotten tomatoes by searching

    one of its bags of props, a database, for a zucchini which will be dealt with in a particular way,

    maybe even in a different language. These actions are all described in the script, which itself

    changes according to what the audience decides to hurtle at the web server, making for an

    interesting play indeed! It is in this way that PHP gives the web server instructions on rendering

    web pages, in this case in JavaScript (http://www.w3schools.com/js/), HTML, and CSS, based

    on data from our databases, which are structured and dealt with in MySQL

    (http://dev.mysql.com/doc/), and POSTGRE/POSTGIS (http://postgis.refractions.net/

    documentation/). KML (http://code.google.com/apis/kml/documentation/) and the Google Maps

    API (http://www.google.com/apis/maps/documentation/) also fit into the categories of

    technologies which manifest specifically through our webpage. These technologies are more

    familiar than you might think: KML has similarities to HTML and the Google Maps API is

    accessed through JavaScript. Think of KML as the language by which the Google Maps API, a

    third party mapping program (think Google Maps) which we have embedded in our web page,

    renders our specific information of interest, such as markers. For those familiar with XML, think

    of KML as Google proprietary version specifically for communicating geographic information.

    Each of these technologies was chosen for particular reasons. PHP is open source, well

    http://dev.mysql.com/doc/http://postgis.refractions.net/http://code.google.com/apis/kml/http://code.google.com/apis/kml/http://postgis.refractions.net/http://dev.mysql.com/doc/
  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    6/16

    dinoride 6

    documented, easy to use, works on multiple platforms, and easily switches between interaction

    with MySQL and POSTGRES. Dinoride also deals with information outside these realms,

    specifically rideshare information coming from Craigslist in RSS format. Craigslist is the

    largest classified advertisement resource in the world, and its free and online. Craigslist is also

    the largest rideshare resource in the United States, measured by potential shares and certainly by

    the range of cities covered. Craigslist makes much of the information on their site available to

    third parties by offering RSS versions of much of the site. RSS is a cousin to XML (and

    implicitly KML) listed above. It is basically a way of expressing information in a structured

    way, similar to HTML, but with a much greater emphasis on hierarchy and relationships. RSS

    allows sites like dinoride to easily include Craigslist information. In the past few years many

    sites have taken advantage of the wealth of data to be found on Craigslist, especially information

    in the RSS format. These applications are commonly termed mashups, because of their mashing

    together of many technologies and data sources. Therefore, Dinoride can properly be included in

    the category of mashup.

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    7/16

    dinoride 7

    This diagram describes the connections between separate scripts and entities (with blue borders). Arrows symbolize

    the flow of data and instructions from one entity to the next. The white box represents a single web host server, withall external elements outside.

    The heart of the application is spatial functionality within POSTGRES/POSTGIS which

    determines desirable share cities, and their distance from a destination, within a certain distance

    from an approximated route. Specifically POSTGRES/POSTGIS draws a straight line from

    origin to destination, draws a buffer around this line, and then finds all city points which intersect

    this buffer. Craigslist (http://www.craigslist.org) share ads are searched for intersecting cities.

    This search specifically looks for share ads with shares traveling in the same direction as the user

    according to word searches, for purposes of example, image the application searching from b

    city to c city, b city and c city being alphabetically between a city (the origin) and d city (the

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    8/16

    dinoride 8

    destination) . The matching cities with eligible shares display, as well as a route line via Google

    Maps which uses its own routing algorithm. You can see the entire process in action below.

    Dinoride interface

    Proposed Modifications

    I have proposed modifications which have all yet to be realized to dinoride along

    three different criteria: output should only include shares which are appropriate for the user

    route, options and structure should meet the most critical user needs, and the interface itself

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    9/16

    dinoride 9

    should be as user-friendly as possible. I must point out that one user may think a certain feature

    sorely needed while another may find that the lack of that feature makes the interface more user

    friendly. This situation introduces the need for careful tradeoffs that I have attempted to make in

    the following section.

    Output should only include shares which are appropriate for the user route

    Inappropriate shares are suggested for three reasons. First, current output occasionally

    includes shares which are not appropriate for a given route, when that route greatly deviates from

    a straight line due to due to the use of Euclidean distance and a straight line object rather than

    geodesic or route distance and route object. A route object must be captured and stored in spatial

    database so that operations can include this variable. It appears that this can be done with the

    Google Maps API route object. Next, some suggested shares deviate from the route too much

    to reasonably include in a route, once a destination has nearly been reached.

    Suggested share 4, at Washington DC would not make The same route with more shares option. The great

    sense in this case, since additional mileage would difference between matches for these two options isoverwhelm the additional shared mileage, especially evident.relative to the entire route.

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    10/16

    dinoride 10

    To resolve this issue, more intelligent matching should be used, especially as the number of

    share results increases. In any event, better matching must take account of the extra mileage

    necessary to meet a share as a proportion of mileage remaining in the route after that node. This

    task can be completed once the route object has been captured. Finally, there is a great degree of

    difference between shares returned on fastest versus most shares criteria. In order to allow

    finer adjustments to these criteria, geometric database objects should be stored under a projected

    spatial reference.

    Options and structure should meet the most critical user needs

    Most existing user options are in the demonstration phase and will only become possible

    through a broader data model. For example, date can only currently be searched for by a single

    number rather than a full month-day-year type date. The data model will be expanded according

    to two strategies, one active and the other passive. The active strategy involves implementation

    of pattern-based parsing and diversification of data searching. Pattern-based parsing will be

    accomplished through string functions, either within the database or by server side script, based

    on an expandable set of rules, determined by careful consideration of share ads. For instance, if

    on is found in a text string followed by a non-spaced group of non-number characters matching

    an entry in a table of possible month names followed by a number, the non-number character

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    11/16

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    12/16

    dinoride 12

    (http://tonto.eia.doe.gov/oog/info/gdu/gasdiesel.asp), as well as the mileage the user will travel

    and the number of passengers. Users should be able to re-route by checking off listings and

    cities from the list view and from the map view. When a city is checked on or off, the route and

    dependent information should change by AJAX data requests and loading.

    Unrealized mockup based on some of the considerations above including: rerouting based on most efficient route,full criteria options, and fuel share suggestions

    Riders could be notified by email when a route becomes complete via a new nodal

    connection that was previously incomplete. It seems there are an almost endless number of

    enhancements to be made, however it is not necessary that Dinoride implement these directly but

    rather could rather implemented by a third party through a Dinoride API, which could then be

    expanded upon ad infinitum. These third parties, presumably running their own sites, would be

    better suited to offer customized functionality for their distinctive user base.

    http://tonto.eia.doe.gov/oog/info/gdu/gasdiesel.asphttp://tonto.eia.doe.gov/oog/info/gdu/gasdiesel.asp
  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    13/16

    dinoride 13

    Business Model Options

    I have considered a number of potential financial arrangements for deploying and

    supporting dinoride. Below I have described three of the more hopeful concepts. The end result

    will likely combine the best elements of each plan.

    Offering premium rideshare listings to the general public would be a straightforward first

    step, providing a needed stream of revenue, and improving the quality of share data. These

    listings would be available to users who sign up for a period membership, which would entitle

    them to post an unlimited amount of premium rideshare listings, include the maximum amount

    of optional criteria, and show up at the top of any rideshare search, with a premium logo, perhaps

    even allowing links to social networking sites for other premium users through APIs from those

    sites.

    Dinoride could be deployed with an institutional or collegiate focus, benefiting from both

    the strengths of an institutional client, the particular niche needs of a university, and the student

    demographic. Universities have the financial resources (expendable money and strong grant

    capturing abilities) and the motives (prestige, good public relations) to invest in this technology

    especially if their peer institutions are doing so. Consider a natural demographic synthesis: most

    users of dinoride are likely to come from a university-aged demographic, since those within this

    group are relatively more flexible, likely to try a novel concept, and comfortable with their age

    cohort; Universities seem like a natural client. Premium or student listings could be offered to

    university clients at a bulk rate. Students can then pick if they only want to ride with other

    students, but would also be able to take advantage of premium listings.

    Finally, making ride share matching capabilities available to third parties such as local

    governments, travel businesses, and non-governmental organizations with fees according to the

    number of matches made available to their site, perhaps through an API which can be used in a

    customized deployment, doesnt just make good business sense, it also appears to be a working

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    14/16

    dinoride 14

    model in place at Hitch, the New Zealand rideshare application mentioned above. Hitch uses an

    affiliate program to allow their search results to be seen by tourist transport agencies and local

    government rideshare initiates alike, presumably for a fee.

    Conclusion

    The rapid growth in availability of structured data -- such as the Craigslist RSS used by

    Dinoride -- has made it possible for the urban planning professional to implement user-focused

    solutions leveraging vast sources of information, hitherto unavailable. Dinoride is unique in the

    tiny but fast growing field of web-based ridesharing for its ability to access this type of

    information, as well as for its implementation of a spatial algorithm better suited to the American

    ridesharing landscape: relatively sparse residential settlement, many alternate routes, few

    available shares, and long distances of travel. Any ridesharing solution must respond to this

    situation, or risk continued irrelevance for the American roadway traveler. Dinoride, by taking

    account of potential routes and potential shares living along those routes, offers such a solution.

    Dinoride will benefit further from refinement of routing and matching rules, expansion of the

    data model, and extension of the user interface. Whether dinoride be the realm of students,

    specialized users such as tourists, or the general public, there is much room for expansion,

    refinement, and excitement in the field of dynamic web-based ridesharing.

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    15/16

    dinoride 15

    References

    Mearns, B (2007). A Client-Side Rideshare Application Described. Retrieved August 15, 2007,

    Web site: http://cpln716.no-ip.org/d/web/ridesharing.htm

  • 8/12/2019 Automatic Route-based Ridesharing on the Web

    16/16

    dinoride 16

    User Interaction Sequence Diagram