Academia.eduAcademia.edu

Figure 16 – uploaded by Adam G Anderson

See full PDFdownloadDownload figure
> Within the Salim-ASsur archive we see here different glimpses of a subset of the network, which depicts only those nodes explicitly involved in the USinalam affair (AKT 6a, 90-111 & 177). The first graph (a: top, left) continues with the subset shown in Fig. 2.6 and includes 87 nodes and 495 edges. The second graph (b: top, right) is an ego- network of USinalam from the unsupervised disambiguation (76 nodes, 3% of the network), arranged by centrality, with node size and color showing degree. The third graph (c: bottom, left) is the subset of nodes from the USinalam affair from the supervised disambiguation in the attestation network, displaying 57 nodes and 410 edges, and the fourth graph (d: bottom, right) is the same set of nodes in a centrality layout, with the number of attestations following the labels. Node color and size are based on eigenvector centrality, as are the labels. Within the Salim-A&gur archive we see a subset of the network, which shows only those nodes explicitly involved in the USinalam affair (AKT 6a, 90-111 & 177). This subset includes 87 nodes and 495 edges.

Figure 2 > Within the Salim-ASsur archive we see here different glimpses of a subset of the network, which depicts only those nodes explicitly involved in the USinalam affair (AKT 6a, 90-111 & 177). The first graph (a: top, left) continues with the subset shown in Fig. 2.6 and includes 87 nodes and 495 edges. The second graph (b: top, right) is an ego- network of USinalam from the unsupervised disambiguation (76 nodes, 3% of the network), arranged by centrality, with node size and color showing degree. The third graph (c: bottom, left) is the subset of nodes from the USinalam affair from the supervised disambiguation in the attestation network, displaying 57 nodes and 410 edges, and the fourth graph (d: bottom, right) is the same set of nodes in a centrality layout, with the number of attestations following the labels. Node color and size are based on eigenvector centrality, as are the labels. Within the Salim-A&gur archive we see a subset of the network, which shows only those nodes explicitly involved in the USinalam affair (AKT 6a, 90-111 & 177). This subset includes 87 nodes and 495 edges.

Related Figures (84)

formula (IF), we find that the ranks of these two PNs are in fact almost identical: with the latter most likely used as a nickname (‘precious’).'*? On the basis of this introductory
formula (IF), we find that the ranks of these two PNs are in fact almost identical: with the latter most likely used as a nickname (‘precious’).'*? On the basis of this introductory
example, which returned the following ranks and samples: Here we see three ranks for Iddin-I8tar in this letter: first has a sample size of 20, which means it WSD) between the ranks the model generated. Take Iddin-I8tar in letter ICK 4 - I 428 fo
example, which returned the following ranks and samples: Here we see three ranks for Iddin-I8tar in this letter: first has a sample size of 20, which means it WSD) between the ranks the model generated. Take Iddin-I8tar in letter ICK 4 - I 428 fo
polarized ranks, for example: is valuable to provide the mean rank in this example, the problem arises when the model returns
polarized ranks, for example: is valuable to provide the mean rank in this example, the problem arises when the model returns
Here we can observe another indication of the sensitivity of the model. Based on this one this text could be one of the main reasons for Imdi-ilum’s low rank:
Here we can observe another indication of the sensitivity of the model. Based on this one this text could be one of the main reasons for Imdi-ilum’s low rank:
second in the IF of the letter, which results in the following ranks distributions: price for my textiles?” Despite her pios and condescending words, Taram-Kub?t still puts herself
second in the IF of the letter, which results in the following ranks distributions: price for my textiles?” Despite her pios and condescending words, Taram-Kub?t still puts herself
Figures Figure 1.t. Lhe Old Assyrian Lablets A tray of Old Assyrian clay cuneiform tablets from the Ankara museum. The tablets shown here (Nos. Kt 90/k 190-230) were published in 1997 (Michel, Cécile, and Paul Garelli. 1997. Tablettes paléo-assyriennes de Kiiltepe, volume I (Kt 90/k). Paris: De Boccard.).
Figures Figure 1.t. Lhe Old Assyrian Lablets A tray of Old Assyrian clay cuneiform tablets from the Ankara museum. The tablets shown here (Nos. Kt 90/k 190-230) were published in 1997 (Michel, Cécile, and Paul Garelli. 1997. Tablettes paléo-assyriennes de Kiiltepe, volume I (Kt 90/k). Paris: De Boccard.).
Figure 1.3.1. Kultepe, Kanesh
Figure 1.3.1. Kultepe, Kanesh
Figure 1.3.3. Hrozny’s plan of the lower town
Figure 1.3.3. Hrozny’s plan of the lower town
Figure 2.1.4. tne southeast quadrant oF Kanesnh (hiertel 214, 52) This plan of the lower town at Kanesh shows the houses of both the Assyrian and Anatolian merchants in the colony. While we do not yet know many of the names of the inhabitants of these hoses, the following numbers are for those which we have been determined based both on the textual record and further excavations by Prof. T. Ozgiic (see Hertel 2014, 44-51). N.B. This figure connects with the plans on figure 4.11 at houses 68 and 69.
Figure 2.1.4. tne southeast quadrant oF Kanesnh (hiertel 214, 52) This plan of the lower town at Kanesh shows the houses of both the Assyrian and Anatolian merchants in the colony. While we do not yet know many of the names of the inhabitants of these hoses, the following numbers are for those which we have been determined based both on the textual record and further excavations by Prof. T. Ozgiic (see Hertel 2014, 44-51). N.B. This figure connects with the plans on figure 4.11 at houses 68 and 69.
aE SS AO OMe LM ae Se EE Ee See The network visualization is from the supervised graph database (2.5) and shows 1,075 nodes for each of the PNs attested in the AKT 6a, vol. 1 (Larsen 2010). The nodes which have been merged have the attestations in parenthesis. For legibility the edges were not visible in this graph. Both node size and color are based on degree. Labels also reflect the degree. Only a subset of the PNs were selected for the visualization. The nodes are ordered alphabetically (counter clockwise) and nodes on the radial axis are ordered by centrality. The complete graph data for this visualization can be found on GitHub using this URL: [https://github.com/admndrsn/Gephi].
aE SS AO OMe LM ae Se EE Ee See The network visualization is from the supervised graph database (2.5) and shows 1,075 nodes for each of the PNs attested in the AKT 6a, vol. 1 (Larsen 2010). The nodes which have been merged have the attestations in parenthesis. For legibility the edges were not visible in this graph. Both node size and color are based on degree. Labels also reflect the degree. Only a subset of the PNs were selected for the visualization. The nodes are ordered alphabetically (counter clockwise) and nodes on the radial axis are ordered by centrality. The complete graph data for this visualization can be found on GitHub using this URL: [https://github.com/admndrsn/Gephi].
Reconstruction based on Larsen (2002) and Ichisar (1981). Broken lines reflect both questionable and proposed additions to the existing family tree. For the textual attestations supporting these proposed additions see Fig. 5.2.7. Chronological venn diagrams of the major family members, who likely possessed a family ‘archive’ or dossiers, attested within the old texts. Although some of these archives and dossiers have yet to be reconstituted, the dates for these entities shown here indicate synchronic nature of these archives.
Reconstruction based on Larsen (2002) and Ichisar (1981). Broken lines reflect both questionable and proposed additions to the existing family tree. For the textual attestations supporting these proposed additions see Fig. 5.2.7. Chronological venn diagrams of the major family members, who likely possessed a family ‘archive’ or dossiers, attested within the old texts. Although some of these archives and dossiers have yet to be reconstituted, the dates for these entities shown here indicate synchronic nature of these archives.
(CCI 5, /a) This figure shows the ranks from text OAA 1, 50 (CCT 3, 7a), measured against all other letters in the old texts, and the corresponding numbers are shown in the chart below, including the role each PN plays in the letter (1-r ‘letter recipient’; l.w. ‘letter writer’ with * indicating who was listed first). The top four ranks provide a sense of the distribution. The sample size (n) is out o1 40. The mean ranks (Mean) are a measurement taken for just this letter, but this can be contrastec by the overall mean rank (Tot. Mean), and weighted standard deviations are listed for both (WSD).
(CCI 5, /a) This figure shows the ranks from text OAA 1, 50 (CCT 3, 7a), measured against all other letters in the old texts, and the corresponding numbers are shown in the chart below, including the role each PN plays in the letter (1-r ‘letter recipient’; l.w. ‘letter writer’ with * indicating who was listed first). The top four ranks provide a sense of the distribution. The sample size (n) is out o1 40. The mean ranks (Mean) are a measurement taken for just this letter, but this can be contrastec by the overall mean rank (Tot. Mean), and weighted standard deviations are listed for both (WSD).
Figure 2.17. Rank comparison of Puzur-Ti’amtum and Zuba The only attested letter for Puzur-Ti’amtum is TC 3, 44, but with a long list of prominent Assyrians (including PaSu-kén and Imdi-ilum) in the introductory formula, his rank is fairly reliable. Zuba, on the other hand, has 8 attestations in the letters. The similarity between the generated rank scores for each sample (esp. the Top rank and 2nd rank), suggests a high probability that these names could refer to the same entity.
Figure 2.17. Rank comparison of Puzur-Ti’amtum and Zuba The only attested letter for Puzur-Ti’amtum is TC 3, 44, but with a long list of prominent Assyrians (including PaSu-kén and Imdi-ilum) in the introductory formula, his rank is fairly reliable. Zuba, on the other hand, has 8 attestations in the letters. The similarity between the generated rank scores for each sample (esp. the Top rank and 2nd rank), suggests a high probability that these names could refer to the same entity.
si 4M \ Jee BEEENSE BEUAEEE O aes 1 ad Above is Ichisar’s reconstruction (Ichisar 1981, 17), and below is Larsen’s reconstructed family tree (Larsen 1982. 219).
si 4M \ Jee BEEENSE BEUAEEE O aes 1 ad Above is Ichisar’s reconstruction (Ichisar 1981, 17), and below is Larsen’s reconstructed family tree (Larsen 1982. 219).
Here we see a representation of a fictive letter written from AS8Sur-idi and AS8ur-nada to Amur-IStar, Alahum and Asgur-taklaku. Because ASSur-idi is mentioned first in the formula, he presumably ranks higher than all the others, but at least explicitly higher than ASSur-nada and Amur-IStar. Among the recipients of the letter, Amur-IStar explicitly outranks Ali-ahum and A8%ur-taklaku, and Ali-ahum outranks A&8ur-taklaku. Whether ASSur-nada outranks Ali-ahum or AS8ur-taklaku is unknown from this formula, thus there is a certain amount of ambiguity between both the sending and receiving ranks of the formula (cf. Bamman, Anderson, Smith 2013, 2). Se a a ea Eee SN Se See See fe A, See, Se cen eee This hierarchical ordering is based on one degree of separation from A8S8ur-nada, taken from the introductory formulae in which ASSur-nada is mentioned. Note that none of the PNs are disambiguated in this graph, which is why Su-I8tar and Alahum have links spanning multiple generations, as both of these PNs were fairly common names, they each represent at least two different entities. For more on this disambiguation, see chapter four §4.3.1- 2.
Here we see a representation of a fictive letter written from AS8Sur-idi and AS8ur-nada to Amur-IStar, Alahum and Asgur-taklaku. Because ASSur-idi is mentioned first in the formula, he presumably ranks higher than all the others, but at least explicitly higher than ASSur-nada and Amur-IStar. Among the recipients of the letter, Amur-IStar explicitly outranks Ali-ahum and A8%ur-taklaku, and Ali-ahum outranks A&8ur-taklaku. Whether ASSur-nada outranks Ali-ahum or AS8ur-taklaku is unknown from this formula, thus there is a certain amount of ambiguity between both the sending and receiving ranks of the formula (cf. Bamman, Anderson, Smith 2013, 2). Se a a ea Eee SN Se See See fe A, See, Se cen eee This hierarchical ordering is based on one degree of separation from A8S8ur-nada, taken from the introductory formulae in which ASSur-nada is mentioned. Note that none of the PNs are disambiguated in this graph, which is why Su-I8tar and Alahum have links spanning multiple generations, as both of these PNs were fairly common names, they each represent at least two different entities. For more on this disambiguation, see chapter four §4.3.1- 2.
i igtt Q& FeFe REED LVVU aMmmay AQ ALLVOULGAULUEE EEIVLUVYVULAN The graph visualizes the PNs recorded in the subset of old texts used in Michel’s study (1991). The two large nodes represent the archival boundaries as they were reconstructed by Michel, those of Innaya son of Elali (top cluster in blue) and Innaya son of Amuraya (bottom cluster in orange). For the color scale of rank distributions, see Fig. 5.4.2. Refer to my GitHub account for the complete network graph, including nodes and edges files (https://github.com/admndrsn/Gephi).
i igtt Q& FeFe REED LVVU aMmmay AQ ALLVOULGAULUEE EEIVLUVYVULAN The graph visualizes the PNs recorded in the subset of old texts used in Michel’s study (1991). The two large nodes represent the archival boundaries as they were reconstructed by Michel, those of Innaya son of Elali (top cluster in blue) and Innaya son of Amuraya (bottom cluster in orange). For the color scale of rank distributions, see Fig. 5.4.2. Refer to my GitHub account for the complete network graph, including nodes and edges files (https://github.com/admndrsn/Gephi).
sa coat — Pte Ye “FoR Ve BESENSE SBEUREEE D9 SB COREBEAW US BER BEBWE COR WEEE The rows are arranged identically to the names listed in Larsen’s study (1982, 222-3). The added columns indicate the top rank, mean rank and the weighted standard deviation (WSD). The ranks in these columns are generated from the I.Fs found on the letters within the Imdi-ilum archive. The mean and WSD are averages measured across all the ranked entries for these PNs in comparison to Imdt1-ilum. For the individual ranks by text see the latest version of the OA annotation network (2.0) in my GitHub account: (https://github.com/admndrsn/Gephi).
sa coat — Pte Ye “FoR Ve BESENSE SBEUREEE D9 SB COREBEAW US BER BEBWE COR WEEE The rows are arranged identically to the names listed in Larsen’s study (1982, 222-3). The added columns indicate the top rank, mean rank and the weighted standard deviation (WSD). The ranks in these columns are generated from the I.Fs found on the letters within the Imdi-ilum archive. The mean and WSD are averages measured across all the ranked entries for these PNs in comparison to Imdt1-ilum. For the individual ranks by text see the latest version of the OA annotation network (2.0) in my GitHub account: (https://github.com/admndrsn/Gephi).
SAUER Soke VII SASSY RARE GQENEILYUD We see here a subset of the attestation network (2.6), which includes the archives we’ve discussed in t im-A8$ur archive as well as the reconstituted ‘archives’ and dossiers in the old texts. It contains 34,061 edges). On the top, left the darker nodes show the 94/k texts found in situ (dark green). On the right are the old texts of unknown provenance (white). On the top, right we see clockwise: the E1ali) dossier in blue, the Imdt1-ilum ‘archive’ in yellow, the AS8ur-nada ‘archive’ in red, the Innaya (s. Amuraya) dossier in teal, and lastly the white texts primarily reflect the PiSu-kén household. Note that the large central node in each of these groups the central node of identical color reflects the ‘archival’ place-holder, whereas t argest node is the actual entity. Also note that these entities are a mix of all the colors because they are more from both the Sa 6,176 nodes (and his study, nnaya (s. he second interconnected to each other (via supervised disambiguation). The bottom row illustrates that the structure of the network is slight y variable while retaining the Gestalt in each iteration.
SAUER Soke VII SASSY RARE GQENEILYUD We see here a subset of the attestation network (2.6), which includes the archives we’ve discussed in t im-A8$ur archive as well as the reconstituted ‘archives’ and dossiers in the old texts. It contains 34,061 edges). On the top, left the darker nodes show the 94/k texts found in situ (dark green). On the right are the old texts of unknown provenance (white). On the top, right we see clockwise: the E1ali) dossier in blue, the Imdt1-ilum ‘archive’ in yellow, the AS8ur-nada ‘archive’ in red, the Innaya (s. Amuraya) dossier in teal, and lastly the white texts primarily reflect the PiSu-kén household. Note that the large central node in each of these groups the central node of identical color reflects the ‘archival’ place-holder, whereas t argest node is the actual entity. Also note that these entities are a mix of all the colors because they are more from both the Sa 6,176 nodes (and his study, nnaya (s. he second interconnected to each other (via supervised disambiguation). The bottom row illustrates that the structure of the network is slight y variable while retaining the Gestalt in each iteration.
Figure 5.3. Old Assyrian family trees
Figure 5.3. Old Assyrian family trees
The chart shows the most central individuals (‘Agents’) found in ca. 2,000 letters. These results combine the network analysis with the latent hierarchical data, generated by Bamman’s model. It immediately speaks to the central importance of the top three individuals across the network. The designator, PN, is used in the OATP database for those instances when a name cannot be identified on a text. It is important to keep in mind that a number of these PNs have not been completely disambiguated, and therefore may represent more than one entity.
The chart shows the most central individuals (‘Agents’) found in ca. 2,000 letters. These results combine the network analysis with the latent hierarchical data, generated by Bamman’s model. It immediately speaks to the central importance of the top three individuals across the network. The designator, PN, is used in the OATP database for those instances when a name cannot be identified on a text. It is important to keep in mind that a number of these PNs have not been completely disambiguated, and therefore may represent more than one entity.
Related topics:
Natural Language ProcessingAssyriologyLatent variable modelingSocial Network Analysis (SNA)Network science
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved