MODIFICATION
A -- Cognitive Information Processing Technology
- Notice Date
- 11/20/2002
- Notice Type
- Modification
- Contracting Office
- Other Defense Agencies, Defense Advanced Research Projects Agency, Contracts Management Office, 3701 North Fairfax Drive, Arlington, VA, 22203-1714
- ZIP Code
- 22203-1714
- Solicitation Number
- BAA02-21
- Response Due
- 6/6/2003
- Archive Date
- 6/6/2004
- Point of Contact
- Ronald Brachman, Information Processing Technology Office (IPTO), Phone none, Fax (703) 741-7804, - Zachary Lemnios, Information Processing Technology Office (IPTO), Phone none, Fax (703) 741-7804,
- E-Mail Address
-
none, none
- Description
- DESCRIPTION: The above referenced BAA is modified to include a new Focal Challenge: "An Enduring, Personalized, Cognitive Assistant (EPCA)." During the life of the BAA, other Focal Challenges may be introduced. The EPCA Focal Challenge is an addition to the areas of interest described in the basic BAA, which include Computational Perception, Representation and Reasoning, Learning, Communications and Interaction Technology, Cognitive Architectures and Integrated Cognitive Agents, Robust Software and Hardware, Cognitive Teams, and Underlying Foundations. The intent of this Focal Challenge is to drive the creation of a new class of integrated, highly functional cognitive systems, central to the technical mission of this BAA and demonstrating the enabling technical elements (reasoning, learning, etc.) for new classes of cognitive information processing applications, as discussed in this BAA. The Enduring Personal Cognitive Assistant will be set in an office context and will act like a personal executive assistant to a modern knowledge-worker or decision-maker. The Assistant should demonstrate a number of key capabilities: continuous learning over significant periods of time (months, if not years), and the ability to survive "injury," intermediate failure and even complete shutdown, whether intentional or inadvertent; this includes the retention of information and skills through system failure and shutdown (Enduring); the capability of autonomous as well as supervised learning, including learning by observing a partner or by being told something directly (at any natural level of abstraction) (Personalized); the capability to have and use domain and task knowledge; the capacity to be aware of events as they transpire and of the Assistant's own place in the world; the ability to have and remember experiences of its own, and to integrate perceptual input with longer-term knowledge; the ability to explain its reasoning and behavior in natural terms to its partner; and the ability to decide what to do and to act in real time (Cognitive); the capability of cooperating in a team or multi-agent situation; the capacity to interact in a multi-modal, broad spectrum way with humans (including natural language); the ability to be available everywhere; and the capacity to develop the degree of trust necessary for successful everyday interaction with a human partner (Assistant). The Assistant must display basic competencies, including interaction with people and other assistant programs in a normal office environment; sending and receiving information in a human-like manner; relating office information and activities contained or described in one medium in a meaningful manner to office information and activities contained or described in another medium (e.g., relating email to calendar entries, phone messages to trips, etc.); and integration with off-the-shelf office technology solutions and applications (e.g., Microsoft Outlook, Netscape Navigator). Our vision is of an Assistant likely to be dedicated to a single person and working intimately with that person over a prolonged period of time (i.e., months). In this long-term setting, the Assistant would observe the behavior of its partner and determine his/her habits and likes and dislikes, for example preferences in travel, information presentation, mail-reading behavior, appointment scheduling, etc. The system must be able to respond to user feedback (either direct or implicit). The Assistant should be able to accept direct natural language direction and advice, ranging, for example, from very specific orders to make a plane reservation to general rules of thumb about scheduling appointments. We would ultimately like the Assistant to be accessible naturally and seamlessly in any environment and modality. The Assistant must be capable of learning and reasoning - for example, learning new office procedures by being trained or directed by a human partner or ultimately, by reading a document describing those procedures. As is a good human assistant, the artificial assistant should be capable of proactively accessing and using electronic information sources locally and on the Internet, for example, to prepare appropriate materials in advance of a meeting. The Assistant must be robust in the face of uncertainty, being able to successfully cope with unanticipated external and possibly disruptive events. The intent of the EPCA Focal challenge is to foster a cross-disciplinary integrated approach that significantly advances the state of the art both at the system level and in core technology. It is essential that technologies and components developed in the course of this research be general enough to lend themselves to other applications (portability to other cognitive applications) - the goal (as is the goal of the entire BAA) is to create powerful and reusable cognitive information processing technologies and techniques rather than simply to create an ad hoc assistant system that serves only as a demonstration. However, successful integration of the cognitive information processing technologies and components into a working overall system is paramount. It is anticipated that there will be multiple awards, which will be reviewed at the 12-month milestone, to include a Preliminary Design Review and demonstration for determining continuation of the technological thrust being reviewed through the next milestones. Proposals submitted in response to the EPCA Focal Challenge must specifically address milestones at the 12-, 24- and 36-month points. At the 12-month milestone, for example, the contractor might address the analysis and development of the initial domain knowledge base; the types of learning algorithms needed to learn procedures and preferences by observation; the specification of the architecture for overall reasoning, decision-making, and learning for the system; and demonstrate elementary meeting scheduling capabilities and conflict resolution, and basic entity and event extraction from email and voicemail transcripts and correlation of such entities and events with calendar entries. Continuing the example, at the 24-month milestone, the contractor might demonstrate simple advice-taking behavior; deeper integration of information across calendars, email, voicemail, etc.; simple preference-inferring behavior; rudimentary modification of information presentation based on stated or inferred user preferences, using confirmatory dialogues; and the initial interaction with other software agents and online sources. At the 36-month milestone, to complete the example, the contractor might demonstrate natural language dialogue-based interaction; proactive information-seeking and preparation of appropriate materials in advance of events needing them (e.g., meetings); general reflective problem-solving to cope with unforeseen events and constraints; and more sophisticated interface behavior around modifications based on inference (e.g., non-annoyance of the user). The establishment of milestones, while at the discretion of the proposer, should clearly provide demonstrable results of the research and integration cumulatively achieved by the team at the milestone described. Proposals must discuss the use of phase or option years, to include fully defining the meaning of phase or option years. Base and phase or option years must be fully priced, to the extent possible. Deliverables, milestones, and demonstrations must be included and clearly defined with links to the Statement of Work. Proposers must propose a multi-organizational but integrated team comprising a Lead System Integration (LSI) function and a set of Technology Focus Centers (TFC's). The LSI function will have overall project management responsibility, to include chief architect and interface control functions, system integration of concepts from the TFC's, and concept validation and evaluation processes. Technology Focus Centers themselves should be multi-organizational, and must reflect a broad and deep representation from the technical community with unique and enabling capabilities for each major technical sub-area of this Focal Challenge; they should participate in the design of all versions of the Assistant system built by the Lead System Integrator, recommend technology elements to the Lead System Integrator, and develop technology elements for all iterations of the architecture and technology concepts for all systems envisioned by the proposal. All proposals must address the inclusion and use of a Technology Transition Committee, comprising DARPA representatives and LSI and TFC representatives. Submitted proposals must clearly discuss the offeror's planned approach to the sharing of information and responsibilities within each Technology Focus Center and between Technology Focus Centers and Lead System Integrators. The proposal must specifically address the proposer and team members' Intellectual Property rights, Government rights to Intellectual Property resulting from the research, and the methods to be used to meaningfully exchange possibly proprietary information among the awardees at the Lead System Integrator level and within the various LSI Integrator and TFC teams. The proposer must specifically address the use of human subjects, if any, at any point during the life of the project. If human subjects are not to be used, a statement of that fact must be included. If human subjects are to be used at any time in the project, the proposer must meet all requirements of Title 45, Code of Federal Regulations (CFR), Part 46 Protection of Human Subjects, effective 13 December 2001, Department of Health and Human Services. The Common Rule (Federal Policy) for the Protection of Human Subjects (56 FR 28003) may also be found at 32 CFR Part 219, Department of Defense. If human subjects are to be used at any time during the project, the proposal must include Industrial Review Board (IRB) approval for the use of human subjects, or a plan for obtaining IRB approval for the use of human subjects prior to the use of the human subjects. Although proposals identified for funding under this Focal Challenge may result in a contract, grant, cooperative agreement, or other transaction depending upon the nature of the work proposed, the required degree of interaction between parties, and other factors, the Government anticipates awarding only contracts in order to maintain the desired level of control over this research. GENERAL INFORMATION: For those proposals submitted directly in response to this modification to BAA 02-21, Focal Challenge: Enduring, Personalized, Cognitive Assistant, a submission date of 12:00 NOON (ET), 19 December 2002 is established for consideration during the initial evaluation phase. Proposal cover sheets, as described in the basic BAA and Proposer's Information Pamphlet, should indicate the EPCA Focal Challenge as the BAA technology area for which the proposal is submitted. Proposals should be submitted to the addresses contained in the basic BAA. The 19 December 2002 submission date applies only to proposals submitted in response to the Focal Challenge. Due to the extremely short submission time, no abstracts responding to the Focal Challenge are required or will be accepted. All other submission requirements, formatting requirements, and reporting requirements remain the same as found in the initial solicitation describing the BAA. Offerors are still encouraged to submit other proposals addressing the issues in the basic BAA and to periodically check the Federal Business Opportunities and the IPTO Solicitation web page for possible new Focal Challenges and other modifications. The full proposal (original and designated number of hard and electronic copies) in response to this Focal Challenge must be submitted in time to reach DARPA by 12:00 noon (ET) Thursday, December 19, 2002, in order to be considered during the initial evaluation phase. However, BAA 02-21 will remain open until 12:00 noon (ET) Friday, June 6, 2003. Thus, proposals for this Focal Challenge may be submitted at any time from issuance of this BAA amendment through Friday, June 6, 2003. While the proposals submitted after the Thursday, December 19, 2002, deadline will be evaluated by the Government, proposers should keep in mind that the likelihood of funding such proposals is significantly less than for those proposals submitted in connection with the initial evaluation and award schedule for this Focal Challenge.
- Record
- SN00208400-W 20021122/021120213543 (fbodaily.com)
- Source
-
FedBizOpps.gov Link to This Notice
(may not be valid after Archive Date)
| FSG Index | This Issue's Index | Today's FBO Daily Index Page |