-尊龙游戏旗舰厅官网
��ࡱ�>�� ������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������u �r�ƍbjbj�n�n2���a��a�d
�������""������������8�d]�2ulyyyyy����s�s�s�s�s�s�s$�v�tyf�s��������s��yy4�t����,�y�y�s���s���y������:ѫ��������su02u��y�f�y��61�y��p����������s�s;p���2u�������������������������������������������������������������������������y���������"qs: risk analysis
volume 42, issue 6, june 2022
1. title: bard: a structured technique for group elicitation of bayesian networks to support analytic reasoning.
authors: nyberg, erik p.; nicholson, ann e.; korb, kevin b.; wybrow, michael; zukerman, ingrid; mascaro, steven; thakur, shreshth; oshni alvandi, abraham; riley, jeff; pearson, ross; morris, shane; herrmann, matthieu; azad, a.k.m.; bolger, fergus; hahn, ulrike; lagnado, david.
abstract: in many complex, real world situations, problem solving and decision making require effective reasoning about causation and uncertainty. however, human reasoning in these cases is prone to confusion and error. bayesian networks (bns) are an artificial intelligence technology that models uncertain situations, supporting better probabilistic and causal reasoning and decision making. however, to date, bn methodologies and software require (but do not include) substantial upfront training, do not provide much guidance on either the model building process or on using the model for reasoning and reporting, and provide no support for building bns collaboratively. here, we contribute a detailed description and motivation for our new methodology and application, bayesian argumentation via delphi (bard). bard utilizes bns and addresses these shortcomings by integrating (1) short, high quality e courses, tips, and help on demand; (2) a stepwise, iterative, and incremental bn construction process; (3) report templates and an automated explanation tool; and (4) a multiuser web based software platform and delphi style social processes. the result is an end to end online platform, with associated online training, for groups without prior bn expertise to understand and analyze a problem, build a model of its underlying probabilistic causal structure, validate and reason with the causal model, and (optionally) use it to produce a written analytic report. initial experiments demonstrate that, for suitable problems, bard aids in reasoning and reporting. comparing their effect sizes also suggests bard's bn building and collaboration combine beneficially and cumulatively.
2. title: a flexible method for parameterizing ranked nodes in bayesian networks using beta distributions.
authors: mascaro, steven; woodberry, owen.
abstract: when novice modelers first attempt to build a bayesian network, they are often impressed with the intuitive graphical structures that capture their causal understanding. this favorable impression evaporates on proceeding to parameterization. conditional probability tables (cpt) require parameters for often hundreds of very similar scenarios and specifying them in the absence of data can be overwhelming. the problem is even more severe when eliciting parameters from experts with limited time. often, there is local structure with fewer parameters that better describes the relationship. such structures include the noisy or, decision trees, and equations. these work well for modelers, but can be an issue for experts and particularly groups of experts. an alternative approach is to elicit only a few cpt rows and interpolate the remainder. this is a promising approach, as it can handle unknown structures and multiple experts, but existing techniques can be limited. here, we present a flexible approach called interbeta for performing cpt interpolation with ordered nodes. in the simplest case, just two cpt rows are needed, but this can be easily augmented with further information. the basic approach assumes input independence, but allows dependencies to be reintroduced as required, and can also be combined with other local structures such as decision trees or equations, leaving the interpolator to fill in the gaps. we explain the interbeta method, describe its capabilities and limitations and how it compares to similar approaches and show how it can trade off elicitation effort against faithfully representing expert understanding.
3. title: balancing the elicitation burden and the richness of expert input when quantifying discrete bayesian networks.
authors: barons, martine j.; mascaro, steven; hanea, anca m.
abstract: structured expert judgment (sej) is a method for obtaining estimates of uncertain quantities from groups of experts in a structured way designed to minimize the pervasive cognitive frailties of unstructured approaches. when the number of quantities required is large, the burden on the groups of experts is heavy, and resource constraints may mean that eliciting all the quantities of interest is impossible. partial elicitations can be complemented with imputation methods for the remaining, unelicited quantities. in the case where the quantities of interest are conditional probability distributions, the natural relationship between the quantities can be exploited to impute missing probabilities. here we test the bayesian intelligence interpolation method and its variations for bayesian network conditional probability tables, called "interbeta." we compare the various outputs of interbeta on two cases where conditional probability tables were elicited from groups of experts. we show that interpolated values are in good agreement with experts' values and give guidance on how interbeta could be used to good effect to reduce expert burden in sej exercises.
4. title: co designing and building an expert elicited non parametric bayesian network model: demonstrating a methodology using a bonamia ostreae spread risk case study.
authors: hanea, anca m.; hilton, zo�; knight, ben; p. robinson, andrew.
abstract: the development and use of probabilistic models, particularly bayesian networks (bn), to support risk based decision making is well established. striking an efficient balance between satisfying model complexity and ease of development requires continuous compromise. codesign, wherein the structural content of the model is developed hand in hand with the experts who will be accountable for the parameter estimates, shows promise, as do so called nonparametric bayesian networks (npbns), which provide a light touch approach to capturing complex relationships among nodes. we describe and demonstrate the process of codesigning, building, quantifying, and validating an npbn model for emerging risks and the consequences of potential management decisions using structured expert judgment (sej). we develop a case study of the local spread of a marine pathogen, namely, bonamia ostreae. the bn was developed through a series of semistructured workshops that incorporated extensive feedback from many experts. the model was then quantified with a combination of field and expert elicited data. the idea protocol for sej was used in its hybrid (remote and face to face) form to elicit information about more than 100 parameters. this article focuses on the modeling and quantification process, the methodological challenges, and the way these were addressed.
5. title: risk analysis frameworks used in biological control and introduction of a novel bayesian network tool.
authors: meurisse, nicolas; marcot, bruce g.; woodberry, owen; barratt, barbara i. p.; todd, jacqui h.
abstract: classical biological control, the introduction of natural enemies to new environments to control unwanted pests or weeds, is, despite numerous successful examples, associated with rising concerns about unwanted environmental impacts such as population decline of nontarget species. recognition of these biosafety risks is globally increasing, and prerelease assessments of biological control agents (bcas) have become more rigorous in many countries. we review the current approaches to risk assessment for bcas as used in australasia, europe, and north america. traditionally, these assessments focus on providing assurance about the specificity of a proposed bca, generally via a list of suitable versus nonsuitable hosts determined through laboratory specificity tests (i.e., by determining the bca's physiological host range). the outcome of interactions of proposed agents in the natural environment can differ from laboratory based predictions. potential nontarget host testing may be incomplete, additional ecological barriers under field conditions may limit encounters between bca and nontargets or reduce attack levels, and bcas could disperse to habitats beyond those used by the target species and adversely affect nontarget species. we advocate for the adoption of more comprehensive, ecologically based, probabilistic risk assessment approaches to bca introductions. an example is provided using a bayesian network that can integrate information on probabilities and uncertainties of a bca to spread and establish in new habitats, interact with nontarget species in these habitats, and eventually negatively impact the populations of these nontarget species. our new model, biocontrol adverse impact probability assessment, aims to be incorporated into a structured decision making framework to support national regulatory authorities.
6. title: adoption of a data driven bayesian belief network investigating organizational factors that influence patient safety.
authors: simsekler, mecit can emre; qazi, abroon.
abstract: medical errors pose high risks to patients. several organizational factors may impact the high rate of medical errors in complex and dynamic healthcare systems. however, limited research is available regarding probabilistic interdependencies between the organizational factors and patient safety errors. to explore this, we adopt a data driven bayesian belief network (bbn) model to represent a class of probabilistic models, using the hospital level aggregate survey data from u.k. hospitals. leveraging the use of probabilistic dependence models and visual features in the bbn model, the results shed new light on relationships existing among eight organizational factors and patient safety errors. with the high prediction capability, the data driven approach results suggest that "health and well being" and "bullying and harassment in the work environment" are the two leading factors influencing the number of reported errors and near misses affecting patient safety. this study provides significant insights to understand organizational factors' role and their relative importance in supporting decision making and safety improvements.
7. title: vine regression with bayes nets: a critical comparison with traditional approaches based on a case study on the effects of breastfeeding on iq.
authors: cooke, roger m.; joe, harry; chang, bo.
abstract: regular vines (r vines) copulas build high dimensional joint densities from arbitrary one dimensional margins and (conditional) bivariate copula densities. vine densities enable the computation of all conditional distributions, though the calculations can be numerically intensive. saturated continuous nonparametric bayes nets (cnpbn) are regular vines. computing regression functions from the vine copula density is termed vine regression. the epicycles of regression�including/excluding covariates, interactions, higher order terms, multicollinearity, model fit, transformations, heteroscedasticity, bias�are dispelled. one simply computes the regressions from the vine copula density. only the question of finding an adequate vine copula remains. vine regression is applied to a data set from the national longitudinal study of youth relating breastfeeding to iq. the expected effects of breastfeeding on iq depend on iq, on the baseline level of breastfeeding, on the duration of additional breastfeeding and on the values of other covariates. a child given two weeks breastfeeding can expect to increase his/her iq by 1.5�2 iq points by adding 10 weeks of breastfeeding, depending on values of other covariates. a child given two years breastfeeding can expect to gain from 0.48�0.65 iq points from 10 additional weeks. adding 10 weeks breastfeeding to each of the 3,179 children in this data set has a net present value $50,700,000 according to the bayes net, compared to $29,000,000 according to the linear regression.
8. title: exploiting the capabilities of bayesian networks for engineering risk assessment: causal reasoning through interventions.
authors:
"%&) ,-/78���ʻʻʩ���wobtm=h
hh
h5�ojqj^jajh
hh
hh�"�hu<�5�ojqj^jh�ud5�ojqj^jo(h�"�h�"�o(&h�"�h�"�5�cjojqj^jajo(h�n�5�cjojqj^jajh
2e5�cjojqj^jaj#h
hh
h5�cjojqj^jajh
h5�cjojqj^jaj#h�"�h�"�5�cjojqj^jaj h$-�5�cjojqj^jajo(#h
hh
h5�cjojqj^jaj,-.�� �������'�#�#�$p%o.p.�.*/�:�:������������������������gd�psgd)w�gd$?�gdto�gd�l$gd%j,gdu<�gd�"�$a$gdt4����� � � �
����������������ᬞ�����uj]�oh:]h�zh�z5�ojqj^jh�zh�zh�l$h�l$5�ojqj^jh�l$5�ojqj^jo(hj�5�ojqjo(hiht4ojqj^jo(h
hhj�ojqj^jo(h
hh
hojqj^jh
hh
hojqj^jo(h
hh
ho(ht45�ojqj^jo(hicy5�ojqj^jh�!dh�!d5�ojqj^jh�!dh�!dh�"�hu<�5�ojqj^jh
hhj�5�ojqj^jaj�������������������������������~qc\n�@qh�zh�z5�ojqj^jhto�hto�5�ojqj^jhto�hto�h�"�h�`�5�ojqj^jhto�5�ojqj^jo(h%j,5�ojqjo(hihu<�ojqj^jo(hih�l$ojqj^jo(h�zh�zojqj^jo(h�zh�zojqj^jh�zh�zh�!dh�!d5�ojqj^jh�l$5�ojqj^jo(h�l$h�l$5�ojqj^jhvi�h�l$5�ojqj^jo(���%&'01�#�#�#�#�#�#�#�#�#�$�$�����������rdvn>1h$?�5�ojqj^jo(h�zh�z5�ojqj^jo(h�zh�zo(h$?�h$?�5�ojqj^jh�"�h%j,5�ojqj^jhs/�5�ojqj^jo(h%j,5�ojqjo(hihanojqj^jh�)hto�ojqj^jo(h�zh�z5�ojqj^jh�zh�zojqj^jh�zh�zhto�5�ojqj^jo(h�!dh�!d5�ojqj^jhto�hto�5�ojqj^jhvi�hto�5�ojqj^jo(�$�$�$�$l%n%p%�%�%�),.m.n.o.p.q.s.y.z.�.������⸪�����vi[mi?h|v�h|v�5�ojqj^jh�psh�ps5�ojqj^jh�"�h)w�5�ojqj^jh�ps5�ojqj^jo(h)w�5�ojqjo(hvi�h%j,ojqj^jo(h�)h$?�ojqj^jo(h|v�h|v�ojqj^jh|v�h|v�ojqj^jo(h|v�h|v�o(h$?�5�ojqj^jo(h�!dh�!d5�ojqj^jh�!dh�!dh$?�h$?�5�ojqj^jhvi�h$?�5�ojqj^jo(�.�.�.�.(/)/*/3/4/h2h4h6h:�:�:�:�:�:�:�:�������������sh[m?7h|v�h|v�o(hr7ahr7a5�ojqj^jh�"�h)w�5�ojqj^jhr7a5�ojqj^jo(h)w�5�ojqjo(h|v�h)w�ojqj^jo(h|v�h|v�ojqj^jo(h|v�h|v�ojqj^jh|v�h|v�h$?�h)w�5�ojqj^jhvi�h)w�5�ojqj^jo(h�ps5�ojqj^jo(h�!dh�!d5�ojqj^jh�!dh�!dh�psh�ps5�ojqj^jhvi�h�ps5�ojqj^j�:�;�;�;�;�;#<$<-<.<.=�d�d�d�d�d���ǹ����yk]od7h|�5�ojqj^jo(h)w�5�ojqjo(hvi�h)w�ojqj^jo(h}onh)w�ojqj^jo(h|v�h|v�ojqj^jo(h|v�h|v�ojqj^jh|v�h|v�h$?�h)w�5�ojqj^jhvi�h)w�5�ojqj^jo(h�!dh�!d5�ojqj^jhvi�hvi�5�ojqj^jhr7ahr7a5�ojqj^jhvi�hr7a5�ojqj^jhr7a5�ojqj^jo(h|v�h|v�5�ojqj^jo(�:�;$<�d�dflf8m9m�m~dbqdq�r�s������:�������������������������������������gd�gd�_gdlz�gd�!dgd�)ggd|�gd)w�gdr7a�d�d�d�dfffffhfjflf~f�f@h6m7m8m9m������㬞㎀xj]oa6h)w�5�ojqjo(hvi�h)w�ojqj^jo(h|v�h)w�ojqj^jo(h|v�h|v�ojqj^jh|v�h|v�ojqj^jo(h|v�h|v�o(h$?�h)w�5�ojqj^jhvi�h)w�5�ojqj^jo(h�!dh�!d5�ojqj^jh�!dh�!dhvi�h|�5�ojqj^jh|�5�ojqj^jo(h|v�h|v�5�ojqj^jh|v�h|v�h|�h|�5�ojqj^jh�"�h)w�5�ojqj^j9m:m&��ŝ�]�6��ҩ��n=���in����a7���qb&>!"�h
[�� �lx�z)��w�v�og�%�ţ|8mă#�/��8��'��=���q�u*��c�`c�w{�d6��/�[��yl���%�� ��n� {�\�o�sm�/�~��i�bcax���/=xe��'5��&\�h�4da���,ye\}�)�l�a�����r�^��*�z���y�v-wowz�n�]��:m�v�=t�c���w��r�b���*i�v�u�r��u����|�nc`�|����*^����pk!
ѐ��'theme/theme/_rels/thememanager.xml.rels��m
�0���wooӻ�&݈э���5
6?$q��
�,.�a��i����c2�1h�:�q��m��@rn��;d�`��o7�g�k(m&$r(.1�r'j��њt���8��v�"��aȼ�h�u}��|�$�b{��p����8�g/]�qasم(����#��l�[������pk-!����[content_types].xmlpk-!�֧��60_rels/.relspk-!ky���theme/theme/thememanager.xmlpk-!���f���theme/theme/theme1.xmlpk-!
ѐ��'� theme/theme/_rels/thememanager.xml.relspk]�
�d����� ����$�.�:�d9mjq��čƍ')* ,-.01ghj�:��ƍ(/i�@� @���������h ��0�( �
����0�( �
��b
�s���� ?������ "'(/ow������78ab��hikl
�������"u~��js������@ghorwy}!"��tu��%0")*1\j��bcef�����������6 @ � � !!""f"o"�"�"i#j#�#�#�#�#�#�#�#�#�#�#:%;%�%�%�&�&
''@(a(k)l)�)�)s d �/�/�/�/�/�/�/�1�1j2k2�3�3�3�3�4�4�5�5(6)697@7j7r7]7d7f7m7�7�7�;�;�;�;�<�<�<�<�<�<�<�<�<�< =====&=(=4=j=r=�>�>�>�>�>�>�>�>�d�d�d�d�d�d�d�d�d�d�d�d�d�d�d�d�d�d�d�d�d�d�d�d "&) ..7������qrr[������effhhq��*����gq�� |���*#-#-#6#�#�#�#�#a(d(d(m(�()/):)1/4/4/>/�/�/�/ 0p6s6s6|6
77�7�7>>>>�>�>�>�>�d�d�d�d��q04e��n[g],-�o�����`<|��o";0ri1�%�rrsui��w�i�={u�,� uz*�m� �r=��-9h�s��eme�.~��
{r�v�4�1l2q �"�iaha��og>�u�tp�`�1_%v��nh{��q��u�3�d]-�z:kkkkn;�q0�<�?*�7��:a��^b� �l{��2h�(�#!�q��m4-cqrrs�4>�p]�sao-�?��yn�b~i�(��)ib��>!]�.�`
rv:g&>!��}n!�l��2��*�tp7�"�0r��\�|�md-%;�blrbr]�a
6ji� ��vj�`�[jo 5j90�ry �:[�
k 3q-r�jx 8i� �� �>�,tp� k�|�2~�
��y��k�
?i��� *�?=��#??*�@�4le�dw"
���z�g�kd�2�^e�-h�x�w��t�bs &�x�-z�y��w�i�>�:-��u
kkkkdw"
q�x�-h�
k^r�
�^�vxb�
�=#�/y�pur ����
��'ff�x-z�d�y�(��)fk{z��fv��-�p�'�e��4�!~q�2h`�?i��be��@��fk-<"��l�q�u��y/�x{��]^�~76=&����`
[� ��rh&<�^��oc!��d{�/�"���[��h�2��d���t��w�8i� (f�tl(����}�t�a9�~�8��7l��4� 8�c�0";�g)n��p7v}g�xd|��&�h�w�a�g�b�u�d\*�)#jb�l�d�
)r�}��w�gl$l��r�qh� v���/xcz.�^m~;�|p%�fzps7l�ps�#���<>)��s��vj��h5h 5� a�l[�x� xcz�be�yfdxl�y��raj>�n�cr%v��h�a9�� !`�90�7��}<�q�}q =�7y]��r�-��.��zf�7��
�e�o[kh,~�hmpm~�mg�=�j��>m�?[%��'w�jo8�l�y2\gqwez ��p]74�`a��m�!�|z��z:�*��b!�g;cq�qq�h�nk^r a�x�t
an�gap�y8��x���s��d���z��j�vzq��t\
~��av'po�_m4�i�w��tkpb�`
rq1��yn�pur �p}��� urep�&>!�z[fxx)!�b!x{�ec!�t�k��oc!�v�>b(p!� ]!�3�l��!nr�q�m�!�j#�g-�!�f�d�[z"�u&z�6'�"�zy�q �"ji�_��m�"� ]!�&�"�)��/�"|z�s\�"=v�g��~#�)9~��k#�� �i3#�#?�=#�i\;�j#p%�sh�#���2k�#�|�#ry �ps�#pb��9�#�sa�d $��yj!4$�`�%�qg$�ep$�"e<@v�$:*�$�[��p�$�]'��?[>�?z]%�`�%vd�i1�%�lyk-,�%�,�=��l�%�z�@�x�%�!0��e&poq�bs &�b*&t�(6=&�'w�rh&2\�qp&nyvc��t�&��&�]'�ep$�tw�'~d:(�r�'�kl��>�'��5�l�'�taz��w(r�@�~d:(�_(it~tl(�x�,�dm�(�t�(�pi�(��]�a#�(�j�`�<>)^e��w)!,�f���)�l�%��)mq{�"*�^�i��le*^l�x�ne*n9��^�*�:8t�� �_�n�[? �v ��o�(s� qh� �`� q�f�� {p���c,@�i��#,�dq[mk, @p�qb,�d&d��,�>m��x�,��>�,{u�,�@�q>,-�m4-rt�h�fk-�a-�r�-�j�n�2@�-�ggc���-x*�y���->o.-dt��t�.��z]�.�x��<}7/d|���y/ds}���/: o=tj�/0�u0{z�o�!0�ran 0gy|�5j90�$�}�;1�0��e�c�0fw2
�z��>2"o�m�29!"e�w�2�f\�h�2n�~�w�3ui�� �3dj�u�3��x@�4�;�4�d5��p�4@l�5h 5�^b^�
-a5�pv5�d5tw�'��5��5md-�b�5i�d��v�5
-a5�a
6�`d�#6ghch�81�6�&7��6�a�r!�6q1���@�6xx�6)1:t�&7�x�\�~74\�7bd9t=r�7��gr
f8c�]��jo8�~�8���8��b�n9?oi�q�9�a:�kx_�s�:kkkk�3�:�!�q�o";�4mjjo;�i\;�%�|c;�q��viu;�ru�;�~7�[u�u�`fr>�>�'�es>hus]i�>r]��v�>lrb�mt?<#iqpz�?!{�?p[5=�tf�?�t�&�?�*��o-�?�mt?�s�?
u|3e�?�-x{�8�?�@}��*.}@��)��-�@�t�`�r�@g��z�@z]%hag;��:a�s6ay8���doahu�(�a�le*�&e�ana��.�an���g�a�a�$,&d]�l�d|1�q�i�dbl#g�>e�\���ez�v�n�e�v ��'�e�#�c�hfs�]�=bdf�^^n��z[f�y���f�"�o w~g� l`�#�g�{9=��g=v�g�o�t�{s
h�-h�pv5p@h�og�ghch'\l��&�h�t�e�hoaq�!~q��y#%�q@gku�nr�q�q��qqg$�r)b;r�,�q��0hr�l��zwr�
{r;1�0��y{r�i�r�� �(u�r�$atxcs�#sw/�=fzps8ts)b;r��ts�#�g3%�s�{�s�;�4��g/t�4>�:8t�gn�bd9t]z[t� �m��td $�`�t�s��,!u�rdi�@gku�u0
g�u���uxcs>�uc;�rxyyv�d�^�v4�o�v�j$w�w�w�k|x�6�x�m�2:3�x�q�y���x�3�:��o�x�#s^l�x\<�jg�xq>,- q�x�}<��x_/,j�zy��t��`-y<myv��q�yv^mc���y<z�&xfk{z�kd
�zhp��[�enx�:[�=�^�`
[1 [�0�[(>y��f\�?\i�(��\�m��x�\~�f<1/�\�s��hus]�7y]�>aqs�]���c�]hv�i���]viu;�]^ph$^�^b^�`.�^�=�^�1_�kx_�a9�ji�_l{�� !`�qb,�� l`�y`� ��t�`�=�}�j�`�qq��u�`�>�`_m�t
a��&na[mk,�9a��q���a��5��>�asq�y�3j�a�zwr�a�t�3b@�z��)ib�u~b1�h��u�b�~#�%;�b@v�$03cd.�b�ggcg-�!�va�c�^����c(�a�u�c�0hr��#�c�j$wvd0q|�`d�d&d�@�m�f�d,-�zn�d�|�#�9!"e2k�#�g e�/e� *^ee����l\e�>e��eme�hgqwe�u
�t�edu�efd��&fji�,\f<}7/��zfs{tq�f3%�s!,�fqx{~�fxx�6�a�f6�f<z�bl#g�^blkqg�j9r��g8|k=vz�g�a-�z�g�-9h�k?oi�u�<mebi�rdi�b*&�kywipz�?�b~i�y{r4�i�`-y�^�ix�u�nf�i�~7djtj�/v,�j�8�?w�j�`�t�=�j�4�q\<�j�dk�g�tkzl!{�?�klv:g��3�l<|�<ie�lj l�^(m��k��d(m�=�t�.�mz*�ma�� �m�hn�y`�gn��!�g)nfr>��^^nv��,mn�kj>�n�_�nh#�n�gv'po�
���"�o�"�o6gp�e�oh��.�o�r��v�omg���o]z[thptf�?(ap(s� @p�[? bep=r�7�c�p�qp&��dq�vzq� �<#iq�!�q� ���qq =��,�q�w�#!�q{s
h�4�q�a�f�l�q|1�q6s�~r�q3q-rie�l��j9r �3�c;�rn[g� �r�-�@�y0�sb(p!��s{fm-dt�y�)1:to
�v�$at�6�x� 0w?0w?0w?0w?�.�t�/e��t(ap�=�t�}�t�o�t�u2�&xph$^�enx[�x,mnf�x&xn
�x�0�<�=.y6�f�� oyrgox*�y�x�u��yd]-sq�y7���yxl�y�t�.��y];yw��y�w��u&z�*n{��tazr=��whkzjx �.tzji��z@�z�.�o���z�`�o��%�z�q�wv{�nh{$,&d�*n{�s�?mq{
hr{��6�jbs{�-x{qx{�@�6m�{�d��0q|�~�vgy|��s�
u|�[|�w)�\�| w~g;�|�g ek�|�.�t�y>�|b�5�ds}xyyv�p}�r>�$�}dm�(�m�}j<�v��=�}jbs{�q�}\
~�=�h,~03c��)9~whkzit~�;�~j{�~0�n�n�~l�s{t�<(>y\*�4e�z �j�{p�4\�7��(k�i����x��m�j�����[�zw�3`2 r� �p
;u
j^
h�n�_ati�wb��.h j<sh�to>s�r#�l$�}&�m'�m(�)%j,�p.�i/�2�z2'4t4.6�6*8;i8cj8�9�<�e=k&>e!@�6@jj@r7a�ec�!d�nd�udvf0i4tj�kul%man�jq�%smkxgy�py�^ro_5a�7ahc
2e�lf�qf�)g) ijmi�ckalthllmnnm}on[r�ps0@ui1vywm8xicy�x}�/�#"�wt�mw��=��=��`�j�nj��9�l�.~�v�=h��j��~�du�� �=i�fj�to�4��m��m��2�)w���lz���c�7�u<�[��t��r��������^�v��\�a�vi�|w�r��$�����$?��
��!�s/����8�*��-��_�$-��f�><����r�|v��c�7"����s�?5�����"�gl�bj�h��-���
��"�w�p�?��n�2����y����e�`-�|��d�d�@��d�d�d�d��������������� �#�$�(�)�/�d�@�
�@��$@��,@��<@�"�t@�,�\@�2�6�x@�>�f��@�d��unknown������������g��.�[x� �times new roman5��symbol3.��.�[x� �arial7.���[ @�verdana;���(�[sosimsuna����$b�cambria math 1��h;"�frl�g�c<
z:#|<
z:#|�-!),.:;?]}���� & 6"0000 00
00000��� ������=�@�\�]�^���([{� 0
000000��;�[����������sdsd 2�q��hp��?�����������������������"�2!xx����v���user
windows (u7b��
�����oh�� '��0`������� ��
(
4@hpx�usernormal.dotmwindows �û�167microsoft office word@ʗ1@�y��\�@��ы�<
z:��
��՜.�� ,��0�
x`t|�������microsoft|#sd
!"#$%&'()* ,-./0123456789:;<=>?@abcdefghijk����mnopqrs����uvwxyz[\]^_`abcdefghijklmnopqrstuvwxyz{|}~������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������root entry�������� �f cѫ���data
������������l1table����t�yworddocument����2�summaryinformation(�������������documentsummaryinformation8���������compobj������������n������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������
���� �fmicrosoft word 97-2003 �ĵ�
msworddocword.document.8�9�q