Jennifer Somali Angeyo | Makerere University (original) (raw)
Uploads
Papers by Jennifer Somali Angeyo
The International IDEA Handbook Series seeks to present comparative analysis, information and ins... more The International IDEA Handbook Series seeks to present comparative analysis, information and insights on a range of democratic institutions and processes. Handbooks are aimed primarily at policy makers, politicians, civil society actors and practitioners in the field. They are also of interest to academia, the democracy assistance community and other bodies.
Applicability of artificial intelligence techniques, in evaluating the influence of the environme... more Applicability of artificial intelligence techniques, in evaluating the influence of the environmental factors in legislative data was found amenable in an earlier study SVM performed to satisfying results with a 21.5 percent error rate for passage of legislation as compared to the performance of ANN at 28 percent error rate and K-NN at 29 percent error rate. These techniques reported both collective influence (ANN, K-NN and SVM) and respective influence (SVM one-against-all classifier). Determining the environmental influences individually or in combination with other factors, could only be measurably achieved using other modeling techniques, despite SVM with probabilistic output of 76 percent outperforming PNN with 71 percent out. A triangulation of both statistical and artificial intelligence modeling techniques in classification is thus proposed for decision making support in legislative drafting, given that computations involving statistical approach correctly predicted up to 98...
The relationship between the actors in the law making process is governed by the doctrine of sepa... more The relationship between the actors in the law making process is governed by the doctrine of separation of powers. Scholarly literature state that each of these actors is interdependent on the other and no single agency is able to exercise complete authority. The doctrine, however, can be extended to enable the three branches to act as checks and balances on each other. The study inquires into how the relationship between and amongst the actors in the legislative process covering both the legislative drafting (not covered under the doctrine of separation of powers) and the lawmaking processes can be modelled without infringing on the doctrine of separation of powers. We analyse the application of artificial intelligent approach to modelling and propose the application of agent technology in modelling this relationship.
Applicability of artificial intelligence techniques, in evaluating the influence of the environme... more Applicability of artificial intelligence techniques, in evaluating the influence of the environmental factors in legislative data was found amenable in an earlier study - SVM performed to satisfying results with a 21.5 percent error rate for passage of legislation as compared to the performance of ANN at 28 percent error rate and K-NN at 29 percent error rate. These techniques reported both collective influence (ANN, K-NN and SVM) and respective influence (SVM one-against-all classifier). Determining the environmental influences - individually or in combination with other factors, could only be measurably achieved using other modeling techniques, despite SVM with probabilistic output of 76 percent outperforming PNN with 71 percent out. A triangulation of both statistical and artificial intelligence modeling techniques in classification is thus proposed for decision making support in legislative drafting, given that computations involving statistical approach correctly predicted up to 98.20 percent and placed economic considerations as the most important factor for the passing of a bill with economic connotations. Other predictions involving political, social, cultural factors did not however, perform as well as the PNN and SVM with probabilistic output.
Abstract This study presents how the legislative knowledge pertinent to the passage and/or cons... more Abstract
This study presents how the legislative knowledge pertinent to the passage and/or
consequently non passage of legislation is captured and processed using artificial
neural network techniques. Qualitative data on the influence the environmental
factors (social, political, economical, cultural and other emerging factors) have
on legislative drafting practices were collected given their role in determining
legislative passage.
Two classifier techniques, artificial neural networks and the nearest neighbour
were trained using 200 legislative data and tested using 100 legislative data and
the performance results were at 0.71 for the nearest neighbour technique and 0.72
for the artificial neural network for passable legislation. Artificial neural network
model is proposed for purposes of this study given its performance and flexibility.
Key Words: Artificial Intelligence, Artificial Neural Network, Nearest
Neighbour Classification, Legislative Drafting, Model, Optimization.
cit.mak.ac.ug/iccir/downloads/ICCIR-12, Aug 2012
The most critical stage in legislative drafting is harmonizing the-government interest and the e... more The most critical stage in legislative drafting is harmonizing the-government interest
and the expectation of the governed. These interests are portrayed as environmental
factors in this study and include the cultural, economic, political and social conditions
of a given society within which the legislation is intended to operate.
Training and testing using 275 legislative data showed 78.2% collective infl uence
by the environmental factors; and using the One-Against-All SVM technique of
classifi cation on 500 legislative data signifi cantly measured the infl uence of each of
the environmental factors on the legislative data analyzed.
Key Words: Artifi cial Intelligence, Evaluation, Classifi cation, Environmental
Factors, Legislative Drafting, Performance, Support Vector Machines (SVMs).
ARPN Journal of Science and Technology, Jan 2013
The relationship between the actors in the law making process is governed by the doctrine of sepa... more The relationship between the actors in the law making process is governed by the doctrine of separation of powers. Scholarly
literature state that each of these actors is interdependent on the other and no single agency is able to exercise complete
authority. The doctrine, however, can be extended to enable the three branches to act as checks and balances on each other.
The study inquires into how the relationship between and amongst the actors in the legislative process covering both the
legislative drafting (not covered under the doctrine of separation of powers) and the lawmaking processes can be modelled without infringing on the doctrine of separation of powers.
We analyse the application of artificial intelligent approach to modelling and propose the application of agent technology in
modelling this relationship.
The International IDEA Handbook Series seeks to present comparative analysis, information and ins... more The International IDEA Handbook Series seeks to present comparative analysis, information and insights on a range of democratic institutions and processes. Handbooks are aimed primarily at policy makers, politicians, civil society actors and practitioners in the field. They are also of interest to academia, the democracy assistance community and other bodies.
Applicability of artificial intelligence techniques, in evaluating the influence of the environme... more Applicability of artificial intelligence techniques, in evaluating the influence of the environmental factors in legislative data was found amenable in an earlier study SVM performed to satisfying results with a 21.5 percent error rate for passage of legislation as compared to the performance of ANN at 28 percent error rate and K-NN at 29 percent error rate. These techniques reported both collective influence (ANN, K-NN and SVM) and respective influence (SVM one-against-all classifier). Determining the environmental influences individually or in combination with other factors, could only be measurably achieved using other modeling techniques, despite SVM with probabilistic output of 76 percent outperforming PNN with 71 percent out. A triangulation of both statistical and artificial intelligence modeling techniques in classification is thus proposed for decision making support in legislative drafting, given that computations involving statistical approach correctly predicted up to 98...
The relationship between the actors in the law making process is governed by the doctrine of sepa... more The relationship between the actors in the law making process is governed by the doctrine of separation of powers. Scholarly literature state that each of these actors is interdependent on the other and no single agency is able to exercise complete authority. The doctrine, however, can be extended to enable the three branches to act as checks and balances on each other. The study inquires into how the relationship between and amongst the actors in the legislative process covering both the legislative drafting (not covered under the doctrine of separation of powers) and the lawmaking processes can be modelled without infringing on the doctrine of separation of powers. We analyse the application of artificial intelligent approach to modelling and propose the application of agent technology in modelling this relationship.
Applicability of artificial intelligence techniques, in evaluating the influence of the environme... more Applicability of artificial intelligence techniques, in evaluating the influence of the environmental factors in legislative data was found amenable in an earlier study - SVM performed to satisfying results with a 21.5 percent error rate for passage of legislation as compared to the performance of ANN at 28 percent error rate and K-NN at 29 percent error rate. These techniques reported both collective influence (ANN, K-NN and SVM) and respective influence (SVM one-against-all classifier). Determining the environmental influences - individually or in combination with other factors, could only be measurably achieved using other modeling techniques, despite SVM with probabilistic output of 76 percent outperforming PNN with 71 percent out. A triangulation of both statistical and artificial intelligence modeling techniques in classification is thus proposed for decision making support in legislative drafting, given that computations involving statistical approach correctly predicted up to 98.20 percent and placed economic considerations as the most important factor for the passing of a bill with economic connotations. Other predictions involving political, social, cultural factors did not however, perform as well as the PNN and SVM with probabilistic output.
Abstract This study presents how the legislative knowledge pertinent to the passage and/or cons... more Abstract
This study presents how the legislative knowledge pertinent to the passage and/or
consequently non passage of legislation is captured and processed using artificial
neural network techniques. Qualitative data on the influence the environmental
factors (social, political, economical, cultural and other emerging factors) have
on legislative drafting practices were collected given their role in determining
legislative passage.
Two classifier techniques, artificial neural networks and the nearest neighbour
were trained using 200 legislative data and tested using 100 legislative data and
the performance results were at 0.71 for the nearest neighbour technique and 0.72
for the artificial neural network for passable legislation. Artificial neural network
model is proposed for purposes of this study given its performance and flexibility.
Key Words: Artificial Intelligence, Artificial Neural Network, Nearest
Neighbour Classification, Legislative Drafting, Model, Optimization.
cit.mak.ac.ug/iccir/downloads/ICCIR-12, Aug 2012
The most critical stage in legislative drafting is harmonizing the-government interest and the e... more The most critical stage in legislative drafting is harmonizing the-government interest
and the expectation of the governed. These interests are portrayed as environmental
factors in this study and include the cultural, economic, political and social conditions
of a given society within which the legislation is intended to operate.
Training and testing using 275 legislative data showed 78.2% collective infl uence
by the environmental factors; and using the One-Against-All SVM technique of
classifi cation on 500 legislative data signifi cantly measured the infl uence of each of
the environmental factors on the legislative data analyzed.
Key Words: Artifi cial Intelligence, Evaluation, Classifi cation, Environmental
Factors, Legislative Drafting, Performance, Support Vector Machines (SVMs).
ARPN Journal of Science and Technology, Jan 2013
The relationship between the actors in the law making process is governed by the doctrine of sepa... more The relationship between the actors in the law making process is governed by the doctrine of separation of powers. Scholarly
literature state that each of these actors is interdependent on the other and no single agency is able to exercise complete
authority. The doctrine, however, can be extended to enable the three branches to act as checks and balances on each other.
The study inquires into how the relationship between and amongst the actors in the legislative process covering both the
legislative drafting (not covered under the doctrine of separation of powers) and the lawmaking processes can be modelled without infringing on the doctrine of separation of powers.
We analyse the application of artificial intelligent approach to modelling and propose the application of agent technology in
modelling this relationship.